diff --git a/.gitea/workflows/ci.yml b/.gitea/workflows/ci.yml
index dca3504..7b0b5c3 100644
--- a/.gitea/workflows/ci.yml
+++ b/.gitea/workflows/ci.yml
@@ -28,8 +28,9 @@ jobs:
         # dependency a failure (CLAWHDF5_REQUIRE_INTEROP below).
         run: |
           apt-get update
-          # cmake builds libz-ng-sys (clawhdf5-format's default `fast-deflate`);
-          # rust:latest does not ship it.
+          # cmake builds libz-ng-sys for the opt-in `fast-deflate` (zlib-ng)
+          # steps in ci-test.sh; rust:latest does not ship it. The default
+          # build (pure-Rust zlib-rs) does not need it.
           apt-get install -y --no-install-recommends python3 python3-venv cmake
           python3 -m venv /opt/interop
           /opt/interop/bin/pip install --no-cache-dir h5py numpy netCDF4 xarray
@@ -75,19 +76,6 @@ jobs:
           command -v rustup >/dev/null || curl -sSf --retry 5 https://sh.rustup.rs | sh -s -- -y --profile minimal --default-toolchain none
           rustup toolchain install stable --profile minimal --component clippy
           echo "$HOME/.cargo/bin" >> "$GITHUB_PATH"
-      - name: Build dependencies
-        # libz-ng-sys (clawhdf5-format's default `fast-deflate`) needs cmake.
-        # In the Docker runner the job is root and can install it; a host
-        # runner cannot, so say what is missing rather than fail inside a
-        # build script.
-        run: |
-          if command -v cmake >/dev/null; then cmake --version | head -1; exit 0; fi
-          if [ "$(id -u)" = 0 ]; then
-            apt-get update -qq && apt-get install -y -qq cmake
-          else
-            echo "::error::cmake is not installed on this runner host (needed by libz-ng-sys)"
-            exit 1
-          fi
       - name: Confirm aarch64
         run: |
           test "$(uname -m)" = aarch64
diff --git a/BENCHMARKS.md b/BENCHMARKS.md
index 1d782e4..969d1ed 100644
--- a/BENCHMARKS.md
+++ b/BENCHMARKS.md
@@ -1185,6 +1185,69 @@ cargo run --release --bin ephemeral_perf
 
 ---
 
+## Deflate backend: zlib-rs vs zlib-ng
+
+Measured 2026-09-23 on tank (AMD Ryzen 7 7800X3D, 8C/16T). The default
+deflate backend is now **zlib-rs**, a pure-Rust port of zlib-ng; zlib-ng (C,
+built with cmake) was the default before and is still available as
+`fast-deflate`. Both builds were compiled once into separate target
+directories and run **alternately, three rounds each**; figures are medians.
+
+```bash
+# zlib-rs (default)
+cargo bench -p clawhdf5-filters --bench deflate_bench
+cargo bench -p clawhdf5-bench --bench h5bench_write --features libhdf5-compare -- '^write_2d_chunked/'
+cargo run --release -p clawhdf5-bench --bin read_harness
+# zlib-ng: add --features fast-deflate (filters) or clawhdf5-format/fast-deflate (bench)
+```
+
+| Workload | zlib-rs | zlib-ng | rs / ng |
+|---|---:|---:|---:|
+| HDF5 chunked write, deflate-6, 512×512 f32 | 1.458 ms | 1.484 ms | 0.98 |
+| HDF5 chunked write, deflate-6, 128×128 f32 | 157.6 µs | 152.8 µs | 1.03 |
+| HDF5 chunked write, deflate-6, 32×32 f32 | 62.9 µs | 60.2 µs | 1.05 |
+| HDF5 read, 64 MB chunked + deflate, full | 64.4 ms | 65.2 ms | 0.99 |
+| HDF5 read, 64×64 window (1 chunk) | 0.18 ms | 0.17 ms | 1.06 |
+| HDF5 read, 512×512 window (4–9 chunks) | 4.10 ms | 4.10 ms | 1.00 |
+| HDF5 read, one row / one column | 1.00 / 2.01 ms | 0.95 / 1.95 ms | 1.05 / 1.03 |
+| Raw inflate, 8 MB f64 | 5.92 ms | 6.06 ms | 0.98 |
+| Raw inflate, 1 MB sine | 82.8 µs | 68.6 µs | 1.21 |
+| Raw deflate-6, 8 MB f64 / 1 MB sine | 92.2 / 2.01 ms | 83.1 / 1.84 ms | 1.11 / 1.09 |
+
+Compressed output is **byte-identical** between the two at levels 1, 6 and 9
+on all three inputs, so files do not change size. libhdf5 1.14.6 took 51.4 ms
+for the 512×512 write in the same session (35× the zlib-rs figure).
+
+On the HDF5 paths zlib-rs is within 6% of zlib-ng everywhere, and ahead on the
+largest write. The raw codec loops show zlib-ng still slightly faster at
+compression (~10%), which chunked writes do not expose because encoding runs
+in parallel across chunks.
+
+**Two findings along the way.** The first measurement had zlib-rs 1.2–1.9×
+slower on single-chunk reads and 3.7× slower on a 1 MB inflate — slower even
+than miniz_oxide. Neither was zlib-rs's fault:
+
+1. **Runtime CPU detection was off.** zlib-rs needs its `std` feature to
+   detect and use SIMD at runtime; flate2 turns it on through its default
+   `runtime_detection` feature, which our `default-features = false` flate2
+   dependency was disabling. With it, a 1 MB inflate goes 282 → 83 µs.
+   `clawhdf5-format/zlib-rs` and `clawhdf5-filters/zlib-rs` now enable it.
+2. **The codec was fed through a 32 KiB buffer.** Both deflate paths used
+   flate2's streaming `read::ZlibDecoder` / `write::ZlibEncoder`. A chunk's
+   decompressed size is known, so they now hand the codec the whole input in
+   one call, into an output buffer sized up front. Worth ~5% on chunked
+   writes and ~10% on zlib-ng's 1 MB inflate. It also closed a hole: the
+   streaming reader returned a truncated stream's bytes without an error, so
+   a truncated chunk read back short; it is now an error.
+
+| 1 MB inflate, same build otherwise | zlib-rs | zlib-ng |
+|---|---:|---:|
+| streaming reader, no runtime detection | 284.1 µs | 76.7 µs |
+| one-shot, no runtime detection | 282.3 µs | 68.5 µs |
+| one-shot + runtime detection (shipped) | **82.8 µs** | **68.6 µs** |
+
+---
+
 ## h5bench-Equivalent I/O Benchmarks
 
 Criterion harness mirroring h5bench serial workloads. clawhdf5 benchmarks dated 2026-07-01;
diff --git a/CHANGELOG.md b/CHANGELOG.md
index cdccb42..5f8dc5b 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -3,6 +3,16 @@
 ## Unreleased
 
 ### Upgrade Notes
+- **The default build no longer compiles any C.** Deflate now defaults to the
+  pure-Rust zlib-rs instead of zlib-ng, so building the core crates needs
+  neither cmake nor a C compiler. Speed on HDF5 reads and writes is within 6%
+  of zlib-ng, and compressed output is byte-identical. To keep zlib-ng, enable
+  `fast-deflate` (on `clawhdf5`, `clawhdf5-format` or `clawhdf5-filters`); it
+  overrides zlib-rs wherever it is on.
+- **A truncated deflate chunk is now an error.** It used to read back short,
+  with no error.
+- **Minimum supported Rust is 1.92**, now declared in every crate's
+  `rust-version` and checked in CI.
 - **New stores use the int8 vector index by default.**
   `MemoryConfig::quantized_index` now defaults to `true`: a quarter of the
   index memory, builds 1.8x (x86-64) and 2.3x (Raspberry Pi 5) faster, and
@@ -13,6 +23,30 @@
   `quantized_index = false`, or pass `create --f32-index` to the CLI, to opt
   out. The CLI's `--quantized-index` is still accepted but is now a no-op.
 
+### Build
+- **Pure-Rust default.** `clawhdf5-format`, `clawhdf5-filters` and the
+  `clawhdf5` facade default to the `zlib-rs` deflate backend; `fast-deflate`
+  (zlib-ng) is opt-in. No crate in the default dependency tree of the core
+  crates compiles C, and `ci-test.sh` now fails if one appears. The facade's
+  `fast-deflate` was on by default and is now off. See `BENCHMARKS.md`,
+  "Deflate backend".
+- `zlib-rs` also enables flate2's `runtime_detection`. Without it zlib-rs has
+  no `std`, cannot detect SIMD at runtime, and inflates 3.5x slower; the
+  workspace builds flate2 with `default-features = false`, which had been
+  switching it off.
+- `rust-version = "1.92"` for the whole workspace (the floor: `wgpu` requires
+  it), and CI checks the workspace on exactly that toolchain.
+- CI keeps zlib-ng building and tested; the arm64 job no longer needs cmake.
+
+### Correctness
+- `clawhdf5-format`: **a truncated deflate chunk read back short, with no
+  error.** The deflate filter used flate2's streaming reader, which returns the
+  bytes it has when the input runs out before the end-of-stream marker. It now
+  decodes in one pass into a buffer sized to the chunk and reports a
+  truncated stream as `DecompressionError`. Same fix in `clawhdf5-filters`,
+  where output longer than the stated size was also silently cut off; it is
+  now an error.
+
 ### Defaults
 - `clawhdf5-agent`: `MemoryConfig::quantized_index` defaults to `true` for new
   stores. The reason it had been off — that int8 search was slower on ARM —
@@ -26,6 +60,10 @@
   knew to ask; it now only ever switches the default off.
 
 ### Performance
+- `clawhdf5-format`, `clawhdf5-filters`: both deflate paths hand the codec the
+  whole chunk in one call, into a buffer allocated once, instead of streaming
+  it through a 32 KiB buffer: about 5% on chunked writes and 10% on zlib-ng's
+  1 MB inflate.
 - `clawhdf5-accel`: **`dot_i8` has aarch64 kernels** — `SDOT` for CPUs with
   the ARMv8.2 dot-product extension (Cortex-A76 and later, Neoverse-N1, every
   Apple Silicon generation) and plain NEON (`vmull_s8` + `vpadalq_s16`) for
diff --git a/CLAUDE.md b/CLAUDE.md
index 8a93702..cb10c9e 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -19,7 +19,7 @@ Cargo workspace with 16 crates under `crates/` (plus `libaec-sys`, an internal F
 | `clawhdf5-agent` | Agent memory, session history, knowledge graph storage |
 | `clawhdf5-gpu` | GPU-accelerated I/O via wgpu (hand-written WGSL compute shaders) |
 | `clawhdf5-accel` | CPU SIMD acceleration path |
-| `clawhdf5-migrate` | Schema migration engine |
+| `clawhdf5-migrate` | SQLite → HDF5 agent-memory migration |
 | `clawhdf5-android` | Android JNI bindings |
 | `clawhdf5-cli` | Command-line interface |
 | `clawhdf5-napi` | Node.js native addon bindings |
@@ -27,7 +27,12 @@ Cargo workspace with 16 crates under `crates/` (plus `libaec-sys`, an internal F
 | `clawhdf5-bench` | Benchmark suite |
 
 ## Key Features
-- Zero-dependency HDF5 read/write (no libhdf5 C library required)
+- Zero-C-dependency HDF5 read/write: no libhdf5, and deflate defaults to
+  pure-Rust zlib-rs (`fast-deflate` opts into zlib-ng, which needs cmake).
+  `ci-test.sh` fails if a C-building crate enters the core crates' default
+  tree. flate2 must keep `runtime_detection` with zlib-rs — without it zlib-rs
+  loses SIMD and inflates 3.5x slower. MSRV is 1.92 (`rust-version`, checked
+  in CI).
 - HNSW vector index for semantic similarity search over agent memories — the
   `clawhdf5-agent` `hnsw` feature is **on by default**, so `hybrid_search` uses
   the approximate `clawhdf5-ann` index for the vector stage (the index mirrors
@@ -82,8 +87,8 @@ Cargo workspace with 16 crates under `crates/` (plus `libaec-sys`, an internal F
   `export` do). An unreadable WAL (torn header, bad magic) is quarantined to
   `<store>.h5.wal.corrupt-<ts>` rather than blocking `open()`; a WAL with an
   unknown *newer* version still fails and is left untouched.
-- `MemoryConfig::compression` uses deflate by default; enable the agent's
-  `zstd` feature to compress embeddings with Zstd instead (links libzstd).
+- `MemoryConfig::compression` is off by default; when on, embeddings are
+  deflate-compressed, or Zstd with the agent's `zstd` feature (links libzstd).
 - `Dataset::verify_provenance()` (clawhdf5 facade, `provenance` feature, on by
   default) recomputes a dataset's SHA-256 and compares it against the
   `_provenance_sha256` attribute written automatically on save when
@@ -128,8 +133,9 @@ cargo test --workspace
 
 Keep workflows free of JavaScript actions (`actions/checkout`, `actions/cache`,
 …): `rust:latest` has no `node`, and not every runner reaches GitHub, where
-they are fetched from. Check out with plain `git` instead. Both jobs need
-`cmake` for `libz-ng-sys` (from `clawhdf5-format`'s default `fast-deflate`).
+they are fetched from. Check out with plain `git` instead. The `test` job
+installs `cmake` for the opt-in `fast-deflate` (zlib-ng) steps; the default
+build needs no C toolchain, so `test-arm64` does not.
 All runners are on `gitea-runner` 3.5.0, from `docker.gitea.com/act_runner`
 — `gitea/act_runner:latest` on Docker Hub is frozen at 0.6.1.
 
diff --git a/Cargo.toml b/Cargo.toml
index 61c47f8..2e1f6fe 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -23,6 +23,9 @@ resolver = "2"
 [workspace.package]
 version = "2.7.0"
 edition = "2024"
+# Oldest toolchain that builds the whole workspace; CI checks it. wgpu (in
+# clawhdf5-gpu) requires 1.92.
+rust-version = "1.92"
 license = "MIT"
 repository = "https://git.redclaw.dev/quantumclaw/clawhdf5"
 
diff --git a/README.md b/README.md
index e235648..58799af 100644
--- a/README.md
+++ b/README.md
@@ -3,24 +3,89 @@
 **The memory layer AI agents deserve. One file. Pure Rust. Zero C dependencies.**
 
 [![License: MIT](https://img.shields.io/badge/license-MIT-blue.svg)](LICENSE)
-[![Rust](https://img.shields.io/badge/rust-1.75%2B-orange.svg)](https://www.rust-lang.org)
-[![Tests](https://img.shields.io/badge/tests-1650%2B%20passing-brightgreen.svg)](#performance)
-[![LongMemEval](https://img.shields.io/badge/LongMemEval%20oracle-Turn--Level%20Hit@5%2084%25%20BM25--only-blue.svg)](BENCHMARKS.md#longmemeval-results)
-[![Footprint](https://img.shields.io/badge/footprint-6.5%20KB%2Frecord-lightgrey.svg)](BENCHMARKS.md#memory-footprint)
+[![Rust](https://img.shields.io/badge/rust-1.92%2B-orange.svg)](https://www.rust-lang.org)
+[![Tests](https://img.shields.io/badge/tests-1850%2B-brightgreen.svg)](#building)
+[![LongMemEval](https://img.shields.io/badge/LongMemEval__s-Turn--Level%20Hit@5%2081.4%25%20hybrid-blue.svg)](BENCHMARKS.md#longmemeval-results)
+[![Footprint](https://img.shields.io/badge/on--disk-1.7%20KB%2Frecord-lightgrey.svg)](BENCHMARKS.md#memory-footprint-1)
 
-ClawHDF5 is a pure-Rust HDF5 implementation combined with a research-grade agent memory engine. It gives AI agents persistent, searchable, cryptographically verifiable memory — all stored in a single portable file.
+ClawHDF5 is a pure-Rust HDF5 implementation combined with a research-grade agent memory engine. It gives AI agents persistent, searchable, integrity-checked memory — all stored in a single portable file.
 
 > **Two things live here:**
 > - **A general-purpose, pure-Rust HDF5 library** — zero C dependencies, NetCDF-4 support, SIMD/GPU acceleration. See the **[Crate Map](#crate-map)** and **[BENCHMARKS.md](BENCHMARKS.md)** for the libhdf5 head-to-head numbers.
 > - **An agent memory layer built on top of it** — vector search, knowledge graph, hippocampal-style consolidation, in `clawhdf5-agent`.
 
-```
-cargo add clawhdf5                          # core HDF5 read/write, no agent layer
-cargo add clawhdf5-agent --features agent   # + agent memory layer
+The crates are not on crates.io yet, so depend on them from git:
+
+```toml
+[dependencies]
+clawhdf5       = { git = "https://git.redclaw.dev/quantumclaw/clawhdf5" }  # core HDF5 read/write
+clawhdf5-agent = { git = "https://git.redclaw.dev/quantumclaw/clawhdf5" }  # + agent memory layer
 ```
 
+> **C dependencies, precisely:** the core crates (`clawhdf5`, `clawhdf5-agent`,
+> `-format`, `-io`, `-filters`, `-ann`, `-accel`, `-netcdf4`, `-cli`) build no C
+> code by default — no libhdf5, and deflate is the pure-Rust
+> [zlib-rs](https://github.com/trifectatechfoundation/zlib-rs), which matches
+> zlib-ng on HDF5 reads and writes and produces byte-identical output
+> ([BENCHMARKS.md § Deflate backend](BENCHMARKS.md#deflate-backend-zlib-rs-vs-zlib-ng)).
+> CI fails if a C-building crate enters their default dependency tree. C comes
+> in only when you ask for it: `fast-deflate` (zlib-ng, needs cmake), `zstd`,
+> `szip`, the BLAS backends, `clawhdf5-migrate` (bundled SQLite) and the
+> Node.js bindings.
+
 > **New here?** Start with the **[Quickstart Guide](docs/QUICKSTART.md)** · See **[Use Cases](docs/USE_CASES.md)** · Read **[Benchmarks](BENCHMARKS.md)**
 
+## What's new (v2.2 → v2.7, and unreleased)
+
+Five releases in September 2026. Details, including upgrade notes and every
+breaking change, are in [CHANGELOG.md](CHANGELOG.md).
+
+**HDF5 correctness (read these if you read files with an earlier release)**
+- **Extensible Array chunk indexes returned wrong data** past the 36th chunk —
+  any dataset with one unlimited dimension. Silent: plausible numbers from the
+  wrong chunks. Fixed in v2.7.0; re-read affected data.
+- Fixed and Extensible Array checksums are now verified, so a corrupt chunk
+  index is `ChecksumMismatch` instead of wrong data (v2.7.0).
+- Compound datatypes written with default libver bounds (plain
+  `h5py.File(path, 'w')`) were mis-parsed; HDF5 2.0 compound v5 and native
+  complex (class 11) types now parse (v2.2.0–v2.3.0).
+- Committed datatypes, fill values, soft links and `H5T_STD_REF` references now
+  read correctly; external links and external raw data are explicit errors;
+  `attrs()` no longer silently drops attributes (v2.3.0–v2.5.0).
+- Datasets indexed by a version-2 B-tree now read (v2.5.0).
+
+**Security and robustness**
+- A crafted file could abort any reader via B-tree v2 recursion or explode it
+  via shared children; both are now fast errors (v2.7.0).
+- Virtual-dataset source paths are confined to the file's directory; chunked
+  reads use overflow-checked sizes and fallible allocation, and the facade
+  writes files atomically (v2.3.0).
+- Agent store: single-writer lock plus `open_read_only`; a crash between
+  checkpoint and WAL truncate no longer duplicates entries; unreadable WALs are
+  quarantined instead of blocking `open()` (v2.3.0).
+
+**Search quality and speed**
+- HNSW neighbour selection now uses the paper's diversity heuristic: recall@10
+  at 100K went from 0.31 to 0.98 (v2.4.0).
+- `hybrid_search` is 79–190× faster than v2.3.0 (p50 0.07 ms at 1K, 4.65 ms at
+  100K). It no longer rebuilds BM25 or rewrites the store per query, and the
+  HNSW graph is persisted (v2.4.0).
+- Default fusion weights are now the measured 0.4 / 0.6 (v2.5.0). Re-ranking had
+  been discarding the retrieval score, costing the OpenClaw backend 40.6pp of
+  Hit@1; fixed in v2.6.0.
+- Selection reads decode only the chunks they touch (a 64×64 window: 105 ms to
+  0.39 ms), and full reads are 1.2–1.9× faster (v2.5.0).
+
+**Memory**
+- A loaded store holds ~30% less (embeddings stored once, v2.6.0), and the
+  int8 HNSW index, **on by default for new stores** (unreleased), brings a
+  100K × 384 store to 1.74× the raw vectors. At equal recall it is also faster
+  than `f32`: 1.63× QPS on AVX2, 1.18× on a Raspberry Pi 5 (NEON `SDOT`).
+
+**Tooling**
+- CI now runs the h5py/netCDF4 interop suites for real (they had been skipping
+  silently) and runs an aarch64 job for the NEON kernels.
+
 ---
 
 ## Why ClawhDF5?
@@ -35,7 +100,7 @@ Every AI agent needs memory. Today that means scattered Markdown files, SQLite d
 | Memory consolidation | Manual pruning | Hippocampal-inspired automatic tiers |
 | Temporal queries | Custom code | Native temporal index (716ns) |
 | Multi-modal | Multiple stores | Unified cross-modal search |
-| Security | Hope for the best | Provenance tracking + anomaly detection |
+| Integrity | Hope for the best | Chained-CRC WAL, checksummed chunk indexes, write-anomaly alerts, opt-in SHA-256 dataset provenance |
 | Portability | Config + DB + files | **One `.h5` file. Copy it anywhere.** |
 
 ---
@@ -58,8 +123,28 @@ Figures below are from an independent reproduction run on a second machine (AMD
 | Sequential read (100K f32) | 23.3 µs | 63.6 µs | **2.7×** |
 | Sequential write (100K f32) | 210 µs | 189 µs | **≈ tie** |
 
+The chunked-write row was re-measured on the same machine on 2026-09-23, after
+the default deflate backend became pure-Rust zlib-rs: 1.46 ms against
+libhdf5's 51.4 ms (**35×**), and 1.48 ms with zlib-ng. libhdf5's own time on
+that machine moved from 65.0 to 51.4 ms between the two dates, which is most
+of the difference from 45×; compare same-day numbers only.
+
 ### Vector Search
 
+**HNSW (the default backend for `hybrid_search`)** — `search_harness`, clustered
+384-dim data, M = 16, ef_construction = 64, recall measured against an exact scan.
+See [BENCHMARKS.md § Search harness](BENCHMARKS.md#search-harness-baseline-v230)
+and [§ Quantising the index copy](BENCHMARKS.md#quantising-the-index-copy-quantized_index):
+
+| N = 100K, ef = 64 | recall@10 | QPS | build |
+|---|---:|---:|---:|
+| `f32` index | 0.9945 | 13 399 | 3.2 s |
+| `i8` index + exact re-score (**default for new stores**) | 0.9940 | **21 848** | **1.8 s** |
+
+Before the v2.4.0 neighbour-selection fix, recall@10 at 100K was 0.31.
+
+**Brute-force and IVF paths** (Criterion, i7-12650H):
+
 | Scale | Flat | IVF (nprobe=10) | IVF-PQ | vs MemX¹ |
 |-------|------|-----------------|--------|----------|
 | 1K | **54 µs** | — | — | — |
@@ -75,7 +160,7 @@ Figures below are from an independent reproduction run on a second machine (AMD
 
 | Operation | Latency | Scale |
 |-----------|---------|-------|
-| Hybrid search (RRF) | **222 µs** | 1K records |
+| Hybrid search (`HDF5Memory::hybrid_search`, p50) | **70 µs** / 0.49 ms / 4.65 ms | 1K / 10K / 100K records |
 | BM25 keyword search | **67 µs** | 1K records |
 | Knowledge graph BFS | **24 µs** | 1K entities |
 | Spreading activation | **17 µs** | 100 entities |
@@ -115,13 +200,17 @@ declaration:
 
 Hybrid is the strongest configuration, which is what running two retrieval stages
 is for. The weights matter more than the stages: a sweep of `vector_weight` from
-0.0 to 1.0 found the long-standing `0.7/0.3` default is **strictly dominated** by
-`0.4/0.6` — better on Hit@1, Hit@5, Hit@10 and MRR at both granularities. Use
-`0.4/0.6`, or `0.3/0.7` if rank-1 precision matters most. See
-[BENCHMARKS.md § Weight sweep](BENCHMARKS.md#longmemeval-results).
+0.0 to 1.0 found the old `0.7/0.3` default is **strictly dominated** by
+`0.4/0.6` — better on Hit@1, Hit@5, Hit@10 and MRR at both granularities. Since
+v2.5.0 `0.4/0.6` is the default (`hybrid::DEFAULT_FUSION`, used by
+`unified_search`, `hybrid_search_with` and the OpenClaw backend); callers that
+pass weights to `hybrid_search` explicitly choose their own. Use `0.3/0.7` if
+rank-1 precision matters most. Reciprocal rank fusion is selectable
+(`hybrid::Fusion::Rrf`) but measured worse than the weighted sum. See
+[BENCHMARKS.md § Weight sweep](BENCHMARKS.md#weight-sweep--full-haystack-n500).
 
-Vector embeddings require `--features embeddings`; without it the vector stage is
-inert and only the BM25 row is produced, which is what every previously published
+The benchmark's vector stage requires `clawhdf5-bench`'s `embeddings` feature
+(real MiniLM embeddings); without it the vector stage is inert and only the BM25 row is produced, which is what every previously published
 number here measured.
 
 On the easier `longmemeval_oracle` variant (evidence sessions only) the same
@@ -146,19 +235,37 @@ retrieval recall reported as QA accuracy typically overstates by 20–30 points.
 
 ### Memory Footprint
 
-| Records | File Size | Bytes/Record | With Compression |
-|---------|-----------|--------------|------------------|
-| 1K | ~6.5 MB | ~6.5 KB | ~2.1 MB (3.1x) |
-| 10K | ~65 MB | ~6.5 KB | ~21 MB (3.1x) |
-| 100K | ~645 MB | ~6.5 KB | ~208 MB (3.1x) |
+**On disk** — 384-dim embeddings, 200-char text
+([BENCHMARKS.md § Memory Footprint](BENCHMARKS.md#memory-footprint-1)):
+
+| Records | File Size | Bytes/Record | Gzip-6 compressed |
+|---------|-----------|--------------|-------------------|
+| 1K | 1.7 MB | 1.8 KB | 277 KB (6.1x) |
+| 10K | 17.0 MB | 1.7 KB | 2.7 MB (6.2x) |
+| 100K | 169.8 MB | 1.7 KB | 26.9 MB (6.2x) |
+
+**In memory** — a store reopened from disk, 384-dim `f32`, measured with a
+counting allocator ([BENCHMARKS.md § Memory footprint](BENCHMARKS.md#memory-footprint)):
+
+| Records | Raw vectors | Reopened, `f32` index | Reopened, `i8` index (default) |
+|---------|-------------|-----------------------|--------------------------------|
+| 1K | 1 MiB | 4 MiB (2.40x) | 2 MiB (1.64x) |
+| 10K | 15 MiB | 44 MiB (3.03x) | 27 MiB (1.81x) |
+| 100K | 146 MiB | 399 MiB (2.72x) | **256 MiB (1.74x)** |
+
+Down from 505 MiB (3.44x) at 100K before v2.6.0, when the cache held every
+embedding twice.
 
 ### Consolidation Efficiency
 
+1,000 records (10 signal + 990 noise), `working_capacity = 100`
+([BENCHMARKS.md § Consolidation Efficiency](BENCHMARKS.md#consolidation-efficiency)):
+
 | Metric | Before | After | Delta |
 |--------|--------|-------|-------|
-| Records in store | 1,000 | ~110 | −89% |
-| Hit@1 recall | ~60% | ~90% | +30% |
-| Search latency | ~2.8 ms | ~0.3 ms | **9x faster** |
+| Records in store | 1,000 | 100 | −90% |
+| Hit@1 recall (signal records) | 100% | 100% | no loss |
+| Search latency | 2.75 ms | 0.31 ms | **8.8x faster** |
 
 **Full benchmark details: [BENCHMARKS.md](BENCHMARKS.md)**
 
@@ -166,74 +273,71 @@ retrieval recall reported as QA accuracy typically overstates by 20–30 points.
 
 ## Agent Memory Architecture
 
-ClawhDF5's agent memory engine implements research from 15+ recent papers on agentic memory systems. It's not a toy — it's the real thing.
+ClawhDF5's agent memory engine draws on 15+ recent papers on agentic memory systems (see [Research Foundation](#research-foundation)).
 
 ```
-                        ┌─────────────────┐
-                        │   Agent Query    │
-                        └────────┬────────┘
-                                 │
-                    ┌────────────▼────────────┐
-                    │   Hybrid Retrieval      │
-                    │  Vector + BM25 + RRF    │
-                    └────────────┬────────────┘
-                                 │
-              ┌──────────────────▼──────────────────┐
-              │         Multi-Factor Re-Ranking      │
-              │  temporal · authority · activation    │
-              └──────────────────┬──────────────────┘
-                                 │
-                    ┌────────────▼────────────┐
-                    │  Confidence Rejection   │
-                    │  (suppress bad matches) │
-                    └────────────┬────────────┘
-                                 │
-        ┌────────────────────────▼────────────────────────┐
-        │              Memory Store (HDF5)                │
-        │                                                 │
-        │  ┌───────────┐ ┌───────────┐ ┌───────────────┐  │
-        │  │ Working   │→│ Episodic  │→│  Semantic     │  │
-        │  │ (bounded) │ │ (bounded) │ │ (long-term)   │  │
-        │  └───────────┘ └───────────┘ └───────────────┘  │
-        │                                                 │
-        │  ┌──────────┐ ┌──────────┐ ┌────────────────┐   │
-        │  │Knowledge │ │Temporal  │ │  Multi-Modal   │   │
-        │  │  Graph   │ │  Index   │ │  Embeddings    │   │
-        │  └──────────┘ └──────────┘ └────────────────┘   │
-        │                                                 │
-        │  ┌──────────┐ ┌──────────┐ ┌────────────────┐   │
-        │  │Provenance│ │ Anomaly  │ │   Source       │   │
-        │  │ Tracking │ │Detection │ │  Isolation     │   │
-        │  └──────────┘ └──────────┘ └────────────────┘   │
-        └─────────────────────────────────────────────────┘
-                              │
-                     ┌────────┴────────┐
-                     │ agent_memory.h5 │
-                     │   single file   │
-                     └─────────────────┘
+                         ┌─────────────────┐
+                         │   Agent Query   │
+                         └────────┬────────┘
+                                  │
+                ┌─────────────────▼──────────────────┐
+                │ HDF5Memory::hybrid_search          │
+                │ HNSW vector + BM25 keyword         │
+                │ weighted fusion (0.4 / 0.6)        │
+                │ × √(Hebbian activation)            │
+                └─────────────────┬──────────────────┘
+                                  │  OpenClaw backend adds:
+                ┌─────────────────▼──────────────────┐
+                │ Multi-factor re-ranking            │
+                │ relevance · recency · authority ·  │
+                │ activation                         │
+                ├────────────────────────────────────┤
+                │ Confidence rejection               │
+                │ (suppress bad matches)             │
+                └─────────────────┬──────────────────┘
+                                  │
+     ┌────────────────────────────▼────────────────────────────┐
+     │ In memory                                               │
+     │  cache (flat f32 embeddings) · BM25 index · HNSW index  │
+     │  provenance ledger + anomaly alerts (session-scoped)    │
+     └────────────────────────────┬────────────────────────────┘
+                                  │ WAL append; checkpoint
+     ┌────────────────────────────▼────────────────────────────┐
+     │ agent_memory.h5   /meta · /memory · /sessions ·         │
+     │                   /knowledge_graph                      │
+     │ agent_memory.h5.wal   chained-CRC write-ahead log       │
+     │ agent_memory.h5.ann   HNSW graph (derived, rebuildable) │
+     │ agent_memory.h5.lock  single-writer lock                │
+     └─────────────────────────────────────────────────────────┘
 ```
 
+Consolidation tiers (Working → Episodic → Semantic), the knowledge-graph
+algorithms, temporal and multi-modal indexes are library components you drive
+directly; the store persists the records, sessions and graph they work over.
+
 ### Module Overview
 
 | Module | What It Does |
 |--------|-------------|
-| **`knowledge`** | Entity/relation graph with BFS traversal, spreading activation, fuzzy entity resolution |
-| **`consolidation`** | Three-tier memory (Working → Episodic → Semantic) with importance scoring and time-decay |
-| **`hybrid`** | Vector + BM25 fusion with Reciprocal Rank Fusion (RRF, k=60). The vector stage uses the HNSW index by default (`hnsw` feature, on by default); disable with `--no-default-features --features float16` for an exact linear scan |
-| **`reranker`** | Multi-factor re-ranking: temporal recency, source authority, activation weight |
-| **`confidence`** | Low-confidence rejection — suppresses spurious recalls when nothing matches |
+| **`knowledge`** | Entity/relation graph with BFS traversal, spreading activation, fuzzy (Levenshtein) entity resolution |
+| **`consolidation`** | Three-tier memory (Working → Episodic → Semantic) with importance scoring, novelty, and time-decay |
+| **`hybrid`** | Vector + BM25 fusion. Default is a min-max-normalised weighted sum, vector 0.4 / keyword 0.6 (`hybrid::DEFAULT_FUSION`, tuned on LongMemEval); RRF is available via `Fusion::Rrf` / `hybrid_search_with`. The vector stage uses the HNSW index by default (`hnsw` feature); disable with `--no-default-features --features float16` for an exact linear scan |
+| **`reranker`** | Multi-factor re-ranking: retrieval relevance (leads, weight 1.0), temporal recency, source authority, activation weight. Used by the OpenClaw backend |
+| **`confidence`** | Low-confidence rejection — suppresses spurious recalls when nothing matches (OpenClaw backend) |
 | **`temporal`** | Sorted timestamp index, session DAG, entity timeline, temporal query hints |
 | **`multimodal`** | Cross-modal search across text/image/audio/video embeddings |
-| **`provenance`** | Source attribution, FNV-1a content hashing, integrity verification |
-| **`anomaly`** | Write rate limiting, 15 injection pattern detectors, source distribution analysis |
+| **`provenance`** | Source attribution and an unkeyed FNV-1a content hash per record, held in memory for the session, for detecting accidental corruption (not tamper-proof) |
+| **`anomaly`** | Write rate limiting, 15 injection-pattern detectors, source-distribution analysis. Alerts never block a save; drain them with `take_anomaly_alerts` |
 | **`openclaw`** | OpenClaw integration: MemoryBackend trait, Markdown ↔ HDF5 conversion |
 | **`vector_search`** | Flat cosine, pre-normed, SIMD, BLAS, GPU, parallel search paths |
-| **`ivf` / `pq`** | IVF-PQ approximate nearest neighbor for billion-scale search |
-| **`bm25`** | BM25 keyword index with TF-IDF scoring |
+| **`ivf` / `pq`** | Standalone IVF and IVF-PQ indexes (benchmarked to 100K vectors); not used by `HDF5Memory`, whose ANN index is HNSW |
+| **`bm25`** | Incremental Okapi BM25 inverted index, kept for the life of the store; optional stemming |
+| **`query_expand`** | Synonym / acronym / temporal query expansion |
 | **`entity_extract`** | Rule-based entity extraction from text chunks into the knowledge graph |
-| **`wal`** | Write-ahead log for crash-safe persistence; each entry is CRC32-checked on replay, so a corrupted entry stops replay there instead of loading bad data |
+| **`wal`** | Write-ahead log (v4) with a chained CRC32 per entry, so a corrupted, reordered, duplicated or spliced entry stops replay; checkpoints record a WAL mark so nothing is applied twice. Appends are not fsynced |
 | **`memory_strategy`** | Pluggable strategies: save-every, semantic-shift, user-correction detection |
 | **`decision_gate`** | Sub-microsecond trivial/substantive classification |
+| **`ephemeral`** | In-memory TTL/LFU working tier |
 | **`async_memory`** | Tokio-based async wrapper over the memory store (`async` feature) |
 
 ---
@@ -265,7 +369,7 @@ assert_eq!(values, vec![22.5, 23.1, 21.8]);
 use clawhdf5_agent::{HDF5Memory, MemoryConfig, MemoryEntry, AgentMemory};
 
 // Create memory store
-let config = MemoryConfig::new("agent.h5", "my-agent", 384);
+let config = MemoryConfig::new("agent.h5".into(), "my-agent", 384);
 let mut memory = HDF5Memory::create(config)?;
 
 // Save a memory
@@ -278,8 +382,8 @@ memory.save(MemoryEntry {
     tags: "preference".into(),
 })?;
 
-// Search
-let results = memory.search(&query_embedding, 5)?;
+// Hybrid search: vector + BM25, weighted 0.4 / 0.6 (the measured default)
+let results = memory.hybrid_search(&query_embedding, "user preferences", 0.4, 0.6, 5);
 for result in results {
     println!("[{:.3}] {}", result.score, result.chunk);
 }
@@ -309,8 +413,8 @@ let neighbors = kg.bfs_neighbors(alice, 2);  // 2-hop neighborhood
 let activated = kg.spreading_activation(&[alice], 0.5, 0.01, 5);
 
 // Entity resolution — fuzzy matching
-let resolved = kg.resolve_or_create("alice", "person", -1, 2);
-// Returns existing Alice entity (Levenshtein distance ≤ 2)
+let (id, created) = kg.resolve_or_create("alice", "person", -1, 2);
+// id == alice, created == false: matched the existing entity (Levenshtein distance ≤ 2)
 ```
 
 ### Memory Consolidation
@@ -321,15 +425,19 @@ use clawhdf5_agent::consolidation::*;
 let config = ConsolidationConfig::default();
 let mut engine = ConsolidationEngine::new(config);
 
-// Add memories — automatically scored for importance
-engine.add_memory("User prefers dark mode", vec![0.1, 0.2, ...], MemorySource::User);
-engine.add_memory("ok", vec![0.0, 0.0, ...], MemorySource::System);
+let now = 1_700_000_000.0; // seconds since the epoch
+
+// Add memories — automatically scored for importance.
+// Elevated sources (System, …) go through a separate, explicit API.
+let id = engine.add_memory("User prefers dark mode".into(), vec![0.1, 0.2, ...], UntrustedSource::User, now);
+engine.add_trusted_memory("ok".into(), vec![0.0, 0.0, ...], TrustedSource::System, now);
 
 // Access a memory (reactivates it)
-engine.access_memory(0);
+engine.access_memory(id, now);
 
 // Run consolidation cycle
-let stats = engine.consolidate();
+engine.consolidate(now);
+let stats = engine.get_stats();
 // Working memories promote to Episodic (if important enough)
 // Episodic memories promote to Semantic (if accessed enough)
 // Low-decay memories get evicted when tiers are full
@@ -357,13 +465,13 @@ let recent = index.latest(10);
 use clawhdf5_agent::openclaw::*;
 
 // Create backend
-let mut backend = ClawhdfBackend::create("memory.h5", "agent-1", 384)?;
+let mut backend = ClawhdfBackend::create(std::path::Path::new("memory.h5"), 384)?;
 
 // Ingest existing Markdown memory files
 let md = std::fs::read_to_string("MEMORY.md")?;
 let count = backend.ingest_markdown("MEMORY.md", &md)?;
 
-// Search (uses full pipeline: RRF → re-rank → confidence filter)
+// Search (full pipeline: weighted vector + BM25 fusion → re-rank → confidence filter)
 let results = backend.search("user preferences", &query_embedding, 5);
 
 // Export back to Markdown
@@ -375,22 +483,23 @@ let exported = backend.export_markdown("MEMORY.md")?;
 ## Crate Map
 
 ```
-clawhdf5 workspace (16 crates, ~92K lines of Rust; plus libaec-sys, an
-                     internal FFI bindings crate for the optional szip feature)
+clawhdf5 workspace (16 crates, ~86K lines of Rust in src/, ~104K with tests
+                     and benches; plus libaec-sys, an internal FFI bindings
+                     crate for the optional szip feature)
 │
 ├── Core HDF5
-│   ├── clawhdf5-format      — Binary parser/writer (no_std), shared type definitions
-│   ├── clawhdf5-io          — I/O abstraction (buffered, mmap, async)
+│   ├── clawhdf5-format      — Binary parser/writer (no_std-capable), shared type definitions
+│   ├── clawhdf5-io          — I/O abstraction (file/memory readers; optional mmap, async, HSDS, MPI)
 │   ├── clawhdf5-filters     — Fast deflate path (zlib-ng); lz4/zstd/pcodec/szip filters live in clawhdf5-format
 │   ├── clawhdf5-derive      — Proc macros
 │   ├── clawhdf5             — High-level API
 │   ├── clawhdf5-netcdf4     — NetCDF-4 support
-│   ├── clawhdf5-accel       — SIMD (NEON, AVX2, AVX-512)
+│   ├── clawhdf5-accel       — SIMD (AVX2, NEON incl. SDOT int8; AVX-512 behind `avx512`)
 │   └── clawhdf5-gpu         — GPU compute (wgpu, hand-written WGSL compute shaders)
 │
 ├── Agent Memory
-│   ├── clawhdf5-agent       — Memory engine (20.9K lines, 32 modules; WAL is CRC32-checked per entry)
-│   ├── clawhdf5-ann         — HNSW approximate nearest neighbor (default backend; optional `parallel` feature)
+│   ├── clawhdf5-agent       — Memory engine (24.7K lines, 32 modules; chained-CRC WAL)
+│   ├── clawhdf5-ann         — HNSW approximate nearest neighbor (default backend; f32 or int8 storage; `parallel` build)
 │   ├── clawhdf5-migrate     — SQLite → HDF5 migration
 │   ├── clawhdf5-android     — Android JNI bridge
 │   └── clawhdf5-cli         — CLI tool
@@ -411,10 +520,10 @@ ClawhDF5's agent memory design draws from 15+ recent papers:
 
 | Paper | Key Insight | ClawhDF5 Module |
 |-------|-------------|-----------------|
-| **MemX** (2026) | RRF + multi-factor re-ranking | `hybrid`, `reranker` |
-| **Graph-Native Cognitive Memory** (2026) | Graph-structured belief revision | `knowledge` |
+| **MemX** (2026) | Hybrid fusion + multi-factor re-ranking | `hybrid`, `reranker` |
+| **Graph-Native Cognitive Memory** (2026) | Graph-structured memory (weighted, timestamped relations; entity timelines) | `knowledge`, `temporal` |
 | **CraniMem** (2026) | Bounded hippocampal memory | `consolidation` |
-| **D-MEM** (2026) | Reward prediction error gating | `consolidation` |
+| **D-MEM** (2026) | Surprise-gated storage (implemented as a novelty score) | `consolidation` |
 | **SYNAPSE** (2025) | Spreading activation for recall | `knowledge` |
 | **RAGdb** (2025) | Zero-dependency edge RAG | Architecture |
 | **MemoryGraft** (2025) | Memory poisoning attacks | `anomaly`, `provenance` |
@@ -429,29 +538,35 @@ ClawhDF5's agent memory design draws from 15+ recent papers:
 
 | Flag | Default | Description |
 |------|---------|-------------|
-| `agent` | no | Full agent memory layer |
-| `float16` | **yes** | Half-precision embedding storage (2× compression) |
+| `float16` | **yes** | Half-precision cosine kernel (`cosine_similarity_f16`). The store itself always writes `f32` embeddings; `MemoryConfig::float16` is recorded in `/meta` but not yet applied |
 | `hnsw` | **yes** | HNSW approximate vector index for `hybrid_search` (via `clawhdf5-ann`); disable for an exact linear scan |
+| `parallel` | **yes** | Parallel HNSW bulk build (same graph, ~3× faster on 16 cores) and Rayon brute-force search strategies |
+| `zstd` | no | Compress embeddings with Zstd instead of deflate when `MemoryConfig::compression` is on (links libzstd) |
+| `fast-math` | no | BLAS matrix-vector multiply |
+| `accelerate` | no | Apple Accelerate / AMX (macOS) |
+| `openblas` | no | OpenBLAS (Linux) |
+| `gpu` | no | GPU search via wgpu |
+| `async` | no | Tokio async with background flush |
+| `agent` | no | Reserved; currently enables nothing (the agent layer is always built) |
+
+To opt out of the parallel build: `--no-default-features --features float16,hnsw`.
+For an exact linear cosine scan instead of HNSW: `--no-default-features --features float16`.
 
 `MemoryConfig::hnsw_m`, `hnsw_ef_construction` and `hnsw_ef_search` tune the
 vector index (16 / 64 / scale-with-`k` by default) and are stored with the
 file.
 
 `MemoryConfig::quantized_index` (**on by default** for new stores) holds the
-HNSW index's own
-copy of the embeddings as `i8`, roughly halving a loaded store's memory
-(2.72x -> 1.74x the raw vectors at 100k x 384). Quantised distances are
-approximate, so the query path re-scores the candidate pool against the exact
-embeddings the store already holds, which keeps recall at the `f32` index's
-level. It is also **faster**: 1.63x the queries per second at equal recall on
-x86-64 (AVX2) and 1.18x on a Raspberry Pi 5 (NEON `SDOT`), with index builds
-1.8x and 2.3x faster respectively. See `BENCHMARKS.md`, "Quantising the index copy".
-| `parallel` | no | Rayon parallel search |
-| `fast-math` | no | BLAS matrix-vector multiply |
-| `accelerate` | no | Apple Accelerate / AMX (macOS) |
-| `openblas` | no | OpenBLAS (Linux) |
-| `gpu` | no | GPU search via wgpu |
-| `async` | no | Tokio async with background flush |
+HNSW index's own copy of the embeddings as `i8`, roughly halving a loaded
+store's memory (2.72x -> 1.74x the raw vectors at 100k x 384). Quantised
+distances are approximate, so the query path re-scores the candidate pool
+against the exact embeddings the store already holds, which keeps recall at the
+`f32` index's level. It is also **faster**: 1.63x the queries per second at
+equal recall on x86-64 (AVX2) and 1.18x on a Raspberry Pi 5 (NEON `SDOT`), with
+index builds 1.8x and 2.3x faster respectively. Stores created before the
+setting existed keep their `f32` index; opt out for new stores with
+`quantized_index = false` or `clawhdf5-cli create --f32-index`. See
+[BENCHMARKS.md § Quantising the index copy](BENCHMARKS.md#quantising-the-index-copy-quantized_index).
 
 ### `clawhdf5-format`
 
@@ -461,26 +576,31 @@ x86-64 (AVX2) and 1.18x on a Raspberry Pi 5 (NEON `SDOT`), with index builds
 | `deflate` | yes | Deflate compression |
 | `checksum` | yes | Jenkins lookup3 verification |
 | `provenance` | yes | SHA-256 provenance attributes |
-| `fast-deflate` | **yes** | zlib-ng backend for faster deflate |
-| `system-zlib-decompress` | **yes** | Use the system zlib for decompression where available |
+| `zlib-rs` | **yes** | Pure-Rust deflate backend ([zlib-rs](https://github.com/trifectatechfoundation/zlib-rs)) |
+| `fast-deflate` | no | zlib-ng deflate backend instead (C; needs `cmake`). Overrides `zlib-rs` when both are on |
+| `system-zlib-decompress` | **yes** | Use Apple's system libz for decompression (macOS only; no effect elsewhere) |
 | `parallel` | no | Parallel chunk encoding + compression (rayon) |
 | `fast-checksum` | no | crc32fast-accelerated checksums |
 | `lz4` | no | LZ4 block compression filter (id 32004) |
 | `zstd` | no | Zstandard compression filter (id 32015) |
 | `pcodec` | no | Pcodec lossless numerical codec (id 32023, via `pco` crate) |
-| `system-zlib` / `zlib-rs` | no | Alternative zlib backends for deflate |
+| `system-zlib` | no | System zlib backend for deflate (C) |
 | `blake3_hash` | no | BLAKE3 content hashing for provenance |
+| `szip` | no | SZIP filter (id 4) via libaec (C, through the internal `libaec-sys` crate) |
 
 ### `clawhdf5-ann`
 
 | Flag | Default | Description |
 |------|---------|-------------|
-| `parallel` | no | Rayon-parallel neighbor-distance computation during HNSW graph pruning |
+| `parallel` | no | Batched bulk build runs neighbour planning and back-link pruning on a Rayon pool; the graph is identical with or without it (enabled by `clawhdf5-agent`'s default `parallel`) |
 
 ### `clawhdf5-io`
 
 | Flag | Default | Description |
 |------|---------|-------------|
+| `mmap` | no | Memory-mapped reads (`memmap2`) |
+| `async` | no | Tokio-based async I/O |
+| `hsds` | no | HSDS (HDF REST service) client |
 | `mpi-io` | no | MPI-backed I/O via the `mpi` crate |
 
 > **Parallel I/O (MPI) limitation:** `mpi-io`'s read path is a root-rank read
@@ -494,17 +614,17 @@ x86-64 (AVX2) and 1.18x on a Raspberry Pi 5 (NEON `SDOT`), with index builds
 ## Building
 
 ```bash
-# Default
+# Default (pure Rust: no cmake or C compiler needed)
 cargo build --workspace
 
 # Agent memory with all accelerations (Linux)
-cargo build -p clawhdf5-agent --features "agent,float16,parallel,fast-math"
+cargo build -p clawhdf5-agent --features fast-math
 
 # Agent memory with Apple Accelerate (macOS)
-cargo build -p clawhdf5-agent --features "agent,float16,accelerate,parallel,gpu"
+cargo build -p clawhdf5-agent --features "accelerate,gpu"
 
 # Tests
-cargo test --workspace            # all 1,650+ tests
+cargo test --workspace            # all 1,850+ tests
 cargo test -p clawhdf5-agent      # agent memory tests
 scripts/ci-test.sh                # what CI runs: fmt, clippy matrix, tests,
                                   # h5py/netCDF4 interop, no_std
@@ -526,25 +646,41 @@ cargo bench -p clawhdf5-bench               # h5bench-equivalent I/O suite
 
 ```
 agent_memory.h5
-├── /meta
-│   ├── schema_version: "1.0"
-│   ├── agent_id, embedder, embedding_dim
-│   └── created_at
+├── /meta                          (attributes)
+│   ├── schema_version: "1.0", edgehdf5_version
+│   ├── agent_id, embedder, embedding_dim, chunk_size, overlap, created_at
+│   ├── float16, compression, compression_level, compact_threshold,
+│   │   hebbian_boost, decay_factor, wal_enabled, wal_max_entries
+│   ├── quantized_index, hnsw_m, hnsw_ef_construction, hnsw_ef_search
+│   ├── wal_applied_len, wal_applied_crc   (WAL mark of the last checkpoint)
+│   └── ann_generation                     (ties the .ann sidecar to this checkpoint)
 ├── /memory
-│   ├── chunks:      string[N]
-│   ├── embeddings:  f32[N × D]  (or f16 with float16 flag)
-│   ├── tombstones:  u8[N]
-│   └── norms:       f32[N]      (pre-computed L2)
+│   ├── chunks:              string[N]
+│   ├── embeddings:          f32[N × D]  (chunked; deflate, or Zstd with the
+│   │                                     `zstd` feature, when compression is on)
+│   ├── source_channel:      string[N]
+│   ├── timestamps:          f64[N]
+│   ├── session_ids:         string[N]
+│   ├── tags:                string[N]
+│   ├── tombstones:          u8[N]
+│   ├── norms:               f32[N]      (pre-computed L2)
+│   └── activation_weights:  f32[N]      (Hebbian)
 ├── /sessions
-│   ├── ids:         string[S]
-│   └── summaries:   string[S]
+│   ├── ids, channels, summaries:  string[S]
+│   ├── start_idxs, end_idxs:      i64[S]
+│   └── timestamps:                f64[S]
 └── /knowledge_graph
-    ├── entity_names:    string[E]
-    ├── relation_srcs:   i64[R]
-    ├── relation_tgts:   i64[R]
-    └── relation_types:  string[R]
+    ├── entity_ids, entity_emb_idxs: i64[E]; entity_names, entity_types: string[E]
+    ├── relation_srcs, relation_tgts: i64[R]; relation_types: string[R]
+    ├── relation_weights: f32[R]; relation_ts: f64[R]
+    └── alias_strings: string[A]; alias_entity_ids: i64[A]   (when aliases exist)
 ```
 
+Alongside the store: `<store>.h5.wal` (write-ahead log), `<store>.h5.ann`
+(HNSW graph; derived, safe to delete) and `<store>.h5.lock` (single-writer
+lock). A second writer gets `MemoryError::Locked`; use
+`HDF5Memory::open_read_only` for a lock-free point-in-time view.
+
 ---
 
 ## Migration
@@ -599,6 +735,6 @@ MIT
 ---
 
 <p align="center">
-  <em>Built by <a href="https://github.com/redclawsystems">RedClaw Systems</a></em><br>
-  <em>~92,000 lines of Rust. Zero C dependencies. One file to remember everything.</em>
+  <em>Built by <a href="https://git.redclaw.dev/quantumclaw">RedClaw Systems</a></em><br>
+  <em>~86,000 lines of Rust. Zero C dependencies. One file to remember everything.</em>
 </p>
diff --git a/crates/clawhdf5-accel/Cargo.toml b/crates/clawhdf5-accel/Cargo.toml
index b24d81d..e59a2ba 100644
--- a/crates/clawhdf5-accel/Cargo.toml
+++ b/crates/clawhdf5-accel/Cargo.toml
@@ -2,6 +2,7 @@
 name = "clawhdf5-accel"
 version = "2.7.0"
 edition = "2024"
+rust-version.workspace = true
 description = "SIMD-accelerated operations for rustyhdf5"
 license = "MIT"
 repository = "https://git.redclaw.dev/quantumclaw/clawhdf5"
diff --git a/crates/clawhdf5-agent/Cargo.toml b/crates/clawhdf5-agent/Cargo.toml
index a7e6796..b8d79ee 100644
--- a/crates/clawhdf5-agent/Cargo.toml
+++ b/crates/clawhdf5-agent/Cargo.toml
@@ -2,6 +2,7 @@
 name = "clawhdf5-agent"
 version = "2.7.0"
 edition = "2024"
+rust-version.workspace = true
 description = "HDF5-backed persistent memory store for on-device AI agents"
 license = "MIT"
 repository = "https://git.redclaw.dev/quantumclaw/clawhdf5"
diff --git a/crates/clawhdf5-android/Cargo.toml b/crates/clawhdf5-android/Cargo.toml
index 2ad609c..38dd2a0 100644
--- a/crates/clawhdf5-android/Cargo.toml
+++ b/crates/clawhdf5-android/Cargo.toml
@@ -2,6 +2,7 @@
 name = "clawhdf5-android"
 version = "2.7.0"
 edition = "2024"
+rust-version.workspace = true
 description = "Android JNI bridge for edgehdf5-memory HDF5 backend"
 license = "MIT"
 
diff --git a/crates/clawhdf5-ann/Cargo.toml b/crates/clawhdf5-ann/Cargo.toml
index 6a5469f..be1b37f 100644
--- a/crates/clawhdf5-ann/Cargo.toml
+++ b/crates/clawhdf5-ann/Cargo.toml
@@ -2,6 +2,7 @@
 name = "clawhdf5-ann"
 version = "2.7.0"
 edition = "2024"
+rust-version.workspace = true
 description = "HNSW approximate nearest neighbor index stored as HDF5"
 license = "MIT"
 repository = "https://git.redclaw.dev/quantumclaw/clawhdf5"
diff --git a/crates/clawhdf5-bench/Cargo.toml b/crates/clawhdf5-bench/Cargo.toml
index 879e921..5ac6350 100644
--- a/crates/clawhdf5-bench/Cargo.toml
+++ b/crates/clawhdf5-bench/Cargo.toml
@@ -2,6 +2,7 @@
 name = "clawhdf5-bench"
 version = "2.7.0"
 edition = "2024"
+rust-version.workspace = true
 description = "Benchmark harnesses for clawhdf5-agent (Track 8)"
 license = "MIT"
 
diff --git a/crates/clawhdf5-cli/Cargo.toml b/crates/clawhdf5-cli/Cargo.toml
index c622e6e..f8e7a15 100644
--- a/crates/clawhdf5-cli/Cargo.toml
+++ b/crates/clawhdf5-cli/Cargo.toml
@@ -2,6 +2,7 @@
 name = "clawhdf5-cli"
 version = "2.7.0"
 edition = "2024"
+rust-version.workspace = true
 license = "MIT"
 description = "CLI for clawhdf5 agent memory — create, save, search, recall, stats"
 repository = "https://git.redclaw.dev/quantumclaw/clawhdf5"
diff --git a/crates/clawhdf5-derive/Cargo.toml b/crates/clawhdf5-derive/Cargo.toml
index e15a46f..c5ca5f1 100644
--- a/crates/clawhdf5-derive/Cargo.toml
+++ b/crates/clawhdf5-derive/Cargo.toml
@@ -2,6 +2,7 @@
 name = "clawhdf5-derive"
 version = "2.7.0"
 edition = "2024"
+rust-version.workspace = true
 description = "Derive macros for rustyhdf5 HDF5 traits"
 license = "MIT"
 repository = "https://git.redclaw.dev/quantumclaw/clawhdf5"
diff --git a/crates/clawhdf5-filters/Cargo.toml b/crates/clawhdf5-filters/Cargo.toml
index a2202b8..dd6060a 100644
--- a/crates/clawhdf5-filters/Cargo.toml
+++ b/crates/clawhdf5-filters/Cargo.toml
@@ -2,6 +2,7 @@
 name = "clawhdf5-filters"
 version = "2.7.0"
 edition = "2024"
+rust-version.workspace = true
 description = "Filter and compression pipeline for clawhdf5"
 license = "MIT"
 repository = "https://git.redclaw.dev/quantumclaw/clawhdf5"
@@ -25,8 +26,12 @@ name = "compression_bench"
 harness = false
 
 [features]
-default = ["fast-deflate"]
+# Pure-Rust zlib-rs by default; `fast-deflate` (zlib-ng, C) overrides it.
+default = ["zlib-rs"]
 fast-deflate = ["flate2/zlib-ng"]
 system-zlib = ["flate2/zlib-default"]
-zlib-rs = ["flate2/zlib-rs"]
+# `runtime_detection` gives zlib-rs `std`, which it needs to detect and use
+# SIMD at runtime. flate2 enables it by default, but we build flate2 with
+# default-features = false, and without it zlib-rs inflates 3.5x slower.
+zlib-rs = ["flate2/zlib-rs", "flate2/runtime_detection"]
 apple-compression = []
diff --git a/crates/clawhdf5-filters/README.md b/crates/clawhdf5-filters/README.md
index 4ccd3fd..45fcf30 100644
--- a/crates/clawhdf5-filters/README.md
+++ b/crates/clawhdf5-filters/README.md
@@ -8,16 +8,18 @@ Filter and compression pipeline for clawhdf5.
 ## Features
 
 - DEFLATE compression/decompression
-- Fast deflate via zlib-ng (`fast-deflate` feature)
+- Pure-Rust deflate via zlib-rs (default, `zlib-rs` feature)
+- zlib-ng instead, if you want it (`fast-deflate` feature; C, needs cmake)
 - Apple Compression framework support (`apple-compression` feature)
 
 ## Usage
 
 ```rust
-use clawhdf5_filters::{deflate_decode, deflate_encode};
+use clawhdf5_filters::{deflate_compress, deflate_decompress};
 
-let compressed = deflate_encode(&data, 6).unwrap();
-let decompressed = deflate_decode(&compressed).unwrap();
+let compressed = deflate_compress(&data, 6).unwrap();
+// The second argument bounds the output: the expected decompressed size.
+let decompressed = deflate_decompress(&compressed, data.len()).unwrap();
 ```
 
 ## License
diff --git a/crates/clawhdf5-filters/src/fast_deflate.rs b/crates/clawhdf5-filters/src/fast_deflate.rs
index 6b4c65a..6ce268f 100644
--- a/crates/clawhdf5-filters/src/fast_deflate.rs
+++ b/crates/clawhdf5-filters/src/fast_deflate.rs
@@ -1,12 +1,13 @@
-//! Fast deflate backends: Apple Compression Framework and zlib-ng.
+//! Deflate backends: Apple Compression Framework, zlib-ng and zlib-rs.
 //!
 //! Backend selection priority (decompression & compression):
 //! 1. Apple Compression Framework (macOS only, `apple-compression` feature)
-//! 2. flate2 with zlib-ng backend (`fast-deflate` feature) or miniz_oxide (default)
+//! 2. flate2 with zlib-ng (`fast-deflate`), else zlib-rs (`zlib-rs`, the
+//!    default), else miniz_oxide
 //!
 //! The Apple Compression Framework uses hardware-accelerated zlib on Apple Silicon
-//! and is typically the fastest option on macOS. zlib-ng is the fastest portable
-//! option and what C HDF5 uses internally.
+//! and is typically the fastest option on macOS. zlib-rs is a pure-Rust port of
+//! zlib-ng; see `BENCHMARKS.md` for how the two compare.
 
 // ---------------------------------------------------------------------------
 // Apple Compression Framework FFI (macOS only)
@@ -243,65 +244,117 @@ mod apple {
 }
 
 // ---------------------------------------------------------------------------
-// Streaming decompression via flate2 (uses zlib-ng when fast-deflate enabled)
+// One-shot (de)compression via flate2 (whichever backend flate2 was built with)
+//
+// The whole input goes to the codec in one call, into an output buffer sized
+// up front. `flate2::read::ZlibDecoder` / `write::ZlibEncoder` stream through a
+// 32 KiB buffer instead, which cost zlib-rs up to 3.7x against zlib-ng on a
+// 1 MB chunk. clawhdf5-format's deflate filter does the same; see
+// `BENCHMARKS.md`, "Deflate backend".
 // ---------------------------------------------------------------------------
 
-/// Streaming decompress with pre-allocated output buffer.
-///
-/// When the output size is known (typical for HDF5 chunks), this avoids
-/// dynamic reallocation by writing directly into a pre-sized buffer.
+/// Decompress into a buffer pre-sized to `output_size`, the expected
+/// decompressed length (known for HDF5 chunks). Output longer than that is an
+/// error, as is a stream that ends early.
 pub(crate) fn flate2_decompress_preallocated(
     data: &[u8],
     output_size: usize,
 ) -> Result<Vec<u8>, String> {
-    use std::io::Read;
-    let mut decoder = flate2::read::ZlibDecoder::new(data);
-    let mut output = vec![0u8; output_size];
-    let mut total_read = 0;
-
-    loop {
-        match decoder.read(&mut output[total_read..]) {
-            Ok(0) => break,
-            Ok(n) => total_read += n,
-            Err(e) => return Err(e.to_string()),
-        }
-    }
-    output.truncate(total_read);
-    Ok(output)
+    inflate_bounded(data, output_size, output_size)
 }
 
 /// Absolute ceiling on decompressed output when the caller has no size hint,
 /// preventing unbounded allocation from a hostile/corrupted zlib stream.
 const MAX_DECOMPRESS_SIZE: usize = 256 * 1024 * 1024;
 
-/// Streaming decompress with dynamic sizing (when output size is unknown).
-///
-/// Bounded by [`MAX_DECOMPRESS_SIZE`] since there is no chunk-size hint to
-/// validate against here — an unbounded `read_to_end` would let a hostile
-/// zlib stream force arbitrarily large allocation (a "zlib bomb").
+/// Decompress with no size hint, bounded by [`MAX_DECOMPRESS_SIZE`] so a
+/// hostile zlib stream cannot force arbitrarily large allocation (a "zlib
+/// bomb").
 pub(crate) fn flate2_decompress_streaming(data: &[u8]) -> Result<Vec<u8>, String> {
-    use std::io::Read;
-    let decoder = flate2::read::ZlibDecoder::new(data);
-    let mut result = Vec::new();
-    decoder
-        .take(MAX_DECOMPRESS_SIZE as u64 + 1)
-        .read_to_end(&mut result)
-        .map_err(|e| e.to_string())?;
-    if result.len() > MAX_DECOMPRESS_SIZE {
-        return Err(format!(
-            "decompressed output exceeds {} MiB limit",
-            MAX_DECOMPRESS_SIZE / 1024 / 1024
-        ));
-    }
-    Ok(result)
+    let hint = data.len().saturating_mul(4).min(1 << 20);
+    inflate_bounded(data, hint, MAX_DECOMPRESS_SIZE).map_err(|e| {
+        if e.ends_with("exceeds size limit") {
+            format!(
+                "decompressed output exceeds {} MiB limit",
+                MAX_DECOMPRESS_SIZE / 1024 / 1024
+            )
+        } else {
+            e
+        }
+    })
 }
 
-/// Compress data using flate2 (zlib-ng when fast-deflate enabled, else miniz_oxide).
+/// Inflate a zlib stream, starting from `size_hint` bytes of output and
+/// failing past `limit`.
+fn inflate_bounded(data: &[u8], size_hint: usize, limit: usize) -> Result<Vec<u8>, String> {
+    use flate2::{Decompress, FlushDecompress, Status};
+
+    // One byte of headroom past the limit distinguishes an over-size stream
+    // from one that legitimately ends exactly at the limit.
+    let max_capacity = limit.saturating_add(1);
+    let mut out = Vec::new();
+    out.try_reserve_exact(size_hint.clamp(1, max_capacity))
+        .map_err(|e| format!("deflate: cannot allocate output: {e}"))?;
+
+    let mut inflater = Decompress::new(true);
+    loop {
+        let (in_before, out_before) = (inflater.total_in(), inflater.total_out());
+        let status = inflater
+            .decompress_vec(
+                &data[in_before as usize..],
+                &mut out,
+                FlushDecompress::Finish,
+            )
+            .map_err(|e| format!("deflate: {e}"))?;
+        if out.len() > limit {
+            return Err("deflate: output exceeds size limit".into());
+        }
+        match status {
+            Status::StreamEnd => return Ok(out),
+            Status::Ok | Status::BufError if out.len() == out.capacity() => {
+                let grow = out.capacity().min(max_capacity - out.capacity()).max(1);
+                out.try_reserve_exact(grow)
+                    .map_err(|e| format!("deflate: cannot allocate output: {e}"))?;
+            }
+            Status::Ok | Status::BufError => {
+                if inflater.total_in() as usize >= data.len()
+                    || (inflater.total_in(), inflater.total_out()) == (in_before, out_before)
+                {
+                    return Err("deflate: truncated stream".into());
+                }
+            }
+        }
+    }
+}
+
+/// Compress data using flate2 (zlib-ng, zlib-rs or miniz_oxide; see module docs).
 pub(crate) fn flate2_compress(data: &[u8], level: u32) -> Result<Vec<u8>, String> {
-    use std::io::Write;
-    let mut encoder = flate2::write::ZlibEncoder::new(Vec::new(), flate2::Compression::new(level));
-    encoder.write_all(data).map_err(|e| e.to_string())?;
-    encoder.finish().map_err(|e| e.to_string())
+    use flate2::{Compress, Compression, FlushCompress, Status};
+
+    // zlib's compressBound, plus the zlib header and trailer.
+    let bound = data.len() + (data.len() >> 12) + (data.len() >> 14) + (data.len() >> 25) + 13 + 6;
+    let mut out = Vec::new();
+    out.try_reserve_exact(bound)
+        .map_err(|e| format!("deflate: cannot allocate output: {e}"))?;
+
+    let mut deflater = Compress::new(Compression::new(level), true);
+    loop {
+        let (in_before, out_before) = (deflater.total_in(), deflater.total_out());
+        let status = deflater
+            .compress_vec(&data[in_before as usize..], &mut out, FlushCompress::Finish)
+            .map_err(|e| format!("deflate: {e}"))?;
+        match status {
+            Status::StreamEnd => return Ok(out),
+            Status::Ok | Status::BufError if out.len() == out.capacity() => out
+                .try_reserve(out.capacity().max(4096))
+                .map_err(|e| format!("deflate: cannot allocate output: {e}"))?,
+            Status::Ok | Status::BufError => {
+                if (deflater.total_in(), deflater.total_out()) == (in_before, out_before) {
+                    return Err("deflate: encoder made no progress".into());
+                }
+            }
+        }
+    }
 }
 
 // ---------------------------------------------------------------------------
@@ -312,7 +365,7 @@ pub(crate) fn flate2_compress(data: &[u8], level: u32) -> Result<Vec<u8>, String
 ///
 /// Selection order:
 /// 1. Apple Compression Framework (macOS + `apple-compression` feature)
-/// 2. flate2 (zlib-ng with `fast-deflate`, otherwise miniz_oxide)
+/// 2. flate2 (zlib-ng with `fast-deflate`, else zlib-rs, else miniz_oxide)
 ///
 /// When `output_hint` > 0, pre-allocates the output buffer for zero-copy
 /// decompression (avoids reallocation).
@@ -344,7 +397,7 @@ pub fn decompress(data: &[u8], output_hint: usize) -> Result<Vec<u8>, String> {
 ///
 /// Selection order:
 /// 1. Apple Compression Framework (macOS + `apple-compression` feature)
-/// 2. flate2 (zlib-ng with `fast-deflate`, otherwise miniz_oxide)
+/// 2. flate2 (zlib-ng with `fast-deflate`, else zlib-rs, else miniz_oxide)
 pub fn compress(data: &[u8], level: u32) -> Result<Vec<u8>, String> {
     #[cfg(all(target_os = "macos", feature = "apple-compression"))]
     {
@@ -377,9 +430,19 @@ pub fn active_backend() -> &'static str {
     {
         "zlib-ng"
     }
+    // flate2 prefers a C zlib over zlib-rs when both are enabled.
+    #[cfg(all(
+        not(all(target_os = "macos", feature = "apple-compression")),
+        not(feature = "fast-deflate"),
+        feature = "zlib-rs"
+    ))]
+    {
+        "zlib-rs"
+    }
     #[cfg(not(any(
         all(target_os = "macos", feature = "apple-compression"),
-        feature = "fast-deflate"
+        feature = "fast-deflate",
+        feature = "zlib-rs"
     )))]
     {
         "miniz_oxide"
@@ -436,7 +499,7 @@ mod tests {
     fn backend_name_is_set() {
         let name = active_backend();
         assert!(
-            ["miniz_oxide", "zlib-ng", "apple-compression"].contains(&name),
+            ["miniz_oxide", "zlib-rs", "zlib-ng", "apple-compression"].contains(&name),
             "unexpected backend: {name}"
         );
     }
diff --git a/crates/clawhdf5-filters/src/lib.rs b/crates/clawhdf5-filters/src/lib.rs
index ae67589..d9d18eb 100644
--- a/crates/clawhdf5-filters/src/lib.rs
+++ b/crates/clawhdf5-filters/src/lib.rs
@@ -2,12 +2,14 @@
 //!
 //! Provides deflate (zlib) decompression/compression with multiple backend options:
 //!
-//! - **Default**: `miniz_oxide` (pure Rust, no C dependencies)
-//! - **`fast-deflate` feature**: `zlib-ng` via flate2 (~2-3x faster, matches C HDF5)
+//! - **Default (`zlib-rs` feature)**: `zlib-rs` via flate2 (pure Rust, no C
+//!   dependencies)
+//! - **`fast-deflate` feature**: `zlib-ng` via flate2 (C, built with cmake)
 //! - **`apple-compression` feature**: Apple Compression Framework on macOS
 //!   (hardware-accelerated on Apple Silicon)
+//! - With none of the above: `miniz_oxide` (pure Rust, slower)
 //!
-//! Backend priority: apple-compression > zlib-ng > miniz_oxide.
+//! Backend priority: apple-compression > zlib-ng > zlib-rs > miniz_oxide.
 
 pub mod fast_deflate;
 
@@ -115,7 +117,7 @@ mod tests {
     fn backend_reports_name() {
         let name = deflate_backend();
         assert!(
-            ["miniz_oxide", "zlib-ng", "apple-compression"].contains(&name),
+            ["miniz_oxide", "zlib-rs", "zlib-ng", "apple-compression"].contains(&name),
             "unexpected backend: {name}"
         );
     }
diff --git a/crates/clawhdf5-format/Cargo.toml b/crates/clawhdf5-format/Cargo.toml
index 7263497..8567a7f 100644
--- a/crates/clawhdf5-format/Cargo.toml
+++ b/crates/clawhdf5-format/Cargo.toml
@@ -2,6 +2,7 @@
 name = "clawhdf5-format"
 version = "2.7.0"
 edition = "2024"
+rust-version.workspace = true
 description = "Pure-Rust HDF5 binary format parsing and writing — no C dependencies"
 license = "MIT"
 repository = "https://git.redclaw.dev/quantumclaw/clawhdf5"
@@ -32,7 +33,10 @@ name = "bench"
 harness = false
 
 [features]
-default = ["std", "checksum", "deflate", "provenance", "fast-deflate", "system-zlib-decompress"]
+# Deflate backend: `zlib-rs` (pure Rust) by default. `fast-deflate` selects
+# zlib-ng instead (C, built with cmake); flate2 prefers a C zlib whenever one
+# is enabled, so turning it on anywhere in the build overrides the default.
+default = ["std", "checksum", "deflate", "provenance", "zlib-rs", "system-zlib-decompress"]
 std = []
 checksum = []
 deflate = ["flate2"]
@@ -42,7 +46,10 @@ fast-checksum = ["crc32fast"]
 fast-deflate = ["flate2/zlib-ng"]
 system-zlib = ["flate2/zlib-default"]
 system-zlib-decompress = []
-zlib-rs = ["flate2/zlib-rs"]
+# `runtime_detection` gives zlib-rs `std`, which it needs to detect and use
+# SIMD at runtime. flate2 enables it by default, but we build flate2 with
+# default-features = false, and without it zlib-rs inflates 3.5x slower.
+zlib-rs = ["flate2/zlib-rs", "flate2/runtime_detection"]
 lz4 = ["lz4_flex"]
 zstd = ["dep:zstd"]
 blake3_hash = ["blake3"]
diff --git a/crates/clawhdf5-format/src/filters.rs b/crates/clawhdf5-format/src/filters.rs
index 9979d3f..c10cea9 100644
--- a/crates/clawhdf5-format/src/filters.rs
+++ b/crates/clawhdf5-format/src/filters.rs
@@ -629,21 +629,70 @@ fn deflate_decompress(data: &[u8], expected_bytes: usize) -> Result<Vec<u8>, For
         // Fall through to flate2 on error
     }
 
-    use std::io::Read;
-    let decoder = flate2::read::ZlibDecoder::new(data);
-    let mut result = Vec::with_capacity(limit.min(1 << 20));
-    // Read one byte past the limit so an over-size stream is distinguishable
+    // A chunk's decompressed size is known, so allocate it once; without one,
+    // start from a multiple of the input and grow.
+    let size_hint = if expected_bytes != 0 {
+        expected_bytes
+    } else {
+        data.len().saturating_mul(4).min(1 << 20)
+    };
+    inflate_bounded(data, size_hint, limit).map_err(FormatError::DecompressionError)
+}
+
+/// Inflate a zlib stream into a buffer sized up front, handing the decoder the
+/// whole input at once.
+///
+/// `flate2::read::ZlibDecoder` feeds its input through a 32 KiB buffer and
+/// grows the output as it goes; on single chunks that cost zlib-rs up to 3.7x
+/// against zlib-ng (`BENCHMARKS.md`, "Deflate backend"). Output beyond `limit`
+/// is an error, as is a stream that ends before its end-of-stream marker (the
+/// streaming reader returned the bytes it had and no error).
+#[cfg(feature = "deflate")]
+pub(crate) fn inflate_bounded(
+    data: &[u8],
+    size_hint: usize,
+    limit: usize,
+) -> Result<Vec<u8>, String> {
+    use flate2::{Decompress, FlushDecompress, Status};
+
+    // One byte of headroom past the limit distinguishes an over-size stream
     // from one that legitimately ends exactly at the limit.
-    decoder
-        .take(limit as u64 + 1)
-        .read_to_end(&mut result)
-        .map_err(|e| FormatError::DecompressionError(e.to_string()))?;
-    if result.len() > limit {
-        return Err(FormatError::DecompressionError(
-            "deflate: output exceeds size limit".into(),
-        ));
+    let max_capacity = limit.saturating_add(1);
+    let mut out = Vec::new();
+    out.try_reserve_exact(size_hint.clamp(1, max_capacity))
+        .map_err(|e| format!("deflate: cannot allocate output: {e}"))?;
+
+    let mut inflater = Decompress::new(true);
+    loop {
+        let (in_before, out_before) = (inflater.total_in(), inflater.total_out());
+        let status = inflater
+            .decompress_vec(
+                &data[in_before as usize..],
+                &mut out,
+                FlushDecompress::Finish,
+            )
+            .map_err(|e| format!("deflate: {e}"))?;
+        if out.len() > limit {
+            return Err("deflate: output exceeds size limit".into());
+        }
+        match status {
+            Status::StreamEnd => return Ok(out),
+            Status::Ok | Status::BufError if out.len() == out.capacity() => {
+                // Out of room: double, up to the limit.
+                let grow = out.capacity().min(max_capacity - out.capacity()).max(1);
+                out.try_reserve_exact(grow)
+                    .map_err(|e| format!("deflate: cannot allocate output: {e}"))?;
+            }
+            Status::Ok | Status::BufError => {
+                // Room left, so the decoder stopped for want of input.
+                if inflater.total_in() as usize >= data.len()
+                    || (inflater.total_in(), inflater.total_out()) == (in_before, out_before)
+                {
+                    return Err("deflate: truncated stream".into());
+                }
+            }
+        }
     }
-    Ok(result)
 }
 
 /// Direct FFI to Apple's system libz for fast decompression.
@@ -722,14 +771,41 @@ fn deflate_decompress(_data: &[u8], _expected_bytes: usize) -> Result<Vec<u8>, F
 /// Compress data with zlib.
 #[cfg(feature = "deflate")]
 fn deflate_compress(data: &[u8], level: u32) -> Result<Vec<u8>, FormatError> {
-    use std::io::Write;
-    let mut encoder = flate2::write::ZlibEncoder::new(Vec::new(), flate2::Compression::new(level));
-    encoder
-        .write_all(data)
-        .map_err(|e| FormatError::CompressionError(e.to_string()))?;
-    encoder
-        .finish()
-        .map_err(|e| FormatError::CompressionError(e.to_string()))
+    deflate_bounded(data, level).map_err(FormatError::CompressionError)
+}
+
+/// Deflate `data` into a zlib stream in one pass, into a buffer sized for the
+/// worst case up front (the same reasoning as [`inflate_bounded`]).
+#[cfg(feature = "deflate")]
+pub(crate) fn deflate_bounded(data: &[u8], level: u32) -> Result<Vec<u8>, String> {
+    use flate2::{Compress, Compression, FlushCompress, Status};
+
+    // zlib's compressBound, plus the zlib header and trailer.
+    let bound = data.len() + (data.len() >> 12) + (data.len() >> 14) + (data.len() >> 25) + 13 + 6;
+    let mut out = Vec::new();
+    out.try_reserve_exact(bound)
+        .map_err(|e| format!("deflate: cannot allocate output: {e}"))?;
+
+    let mut deflater = Compress::new(Compression::new(level), true);
+    loop {
+        let (in_before, out_before) = (deflater.total_in(), deflater.total_out());
+        let status = deflater
+            .compress_vec(&data[in_before as usize..], &mut out, FlushCompress::Finish)
+            .map_err(|e| format!("deflate: {e}"))?;
+        match status {
+            Status::StreamEnd => return Ok(out),
+            // The bound should make running out of room unreachable; grow
+            // rather than fail if it happens.
+            Status::Ok | Status::BufError if out.len() == out.capacity() => out
+                .try_reserve(out.capacity().max(4096))
+                .map_err(|e| format!("deflate: cannot allocate output: {e}"))?,
+            Status::Ok | Status::BufError => {
+                if (deflater.total_in(), deflater.total_out()) == (in_before, out_before) {
+                    return Err("deflate: encoder made no progress".into());
+                }
+            }
+        }
+    }
 }
 
 #[cfg(not(feature = "deflate"))]
@@ -1833,6 +1909,74 @@ mod tests {
         assert!(deflate_decompress(&compressed, 64).is_err());
     }
 
+    #[cfg(feature = "deflate")]
+    fn noisy_bytes(n: usize) -> Vec<u8> {
+        // Compressible but not trivially so.
+        (0..n)
+            .map(|i| ((i as f64 * 0.01).sin() * 127.0 + 128.0) as u8 ^ (i as u8 & 3))
+            .collect()
+    }
+
+    #[test]
+    #[cfg(feature = "deflate")]
+    fn deflate_decompress_accepts_output_exactly_at_chunk_size() {
+        let data = noisy_bytes(100_000);
+        let compressed = deflate_compress(&data, 6).unwrap();
+        assert_eq!(deflate_decompress(&compressed, data.len()).unwrap(), data);
+        // One byte short of the real size is over the limit.
+        assert!(deflate_decompress(&compressed, data.len() - 1).is_err());
+    }
+
+    #[test]
+    #[cfg(feature = "deflate")]
+    fn deflate_decompress_without_size_grows_the_buffer() {
+        // No chunk size: the output starts at 4x the input and has to grow.
+        let data = vec![7u8; 3 * 1024 * 1024];
+        let compressed = deflate_compress(&data, 6).unwrap();
+        assert!(compressed.len() * 4 < data.len());
+        assert_eq!(deflate_decompress(&compressed, 0).unwrap(), data);
+    }
+
+    #[test]
+    #[cfg(feature = "deflate")]
+    fn deflate_decompress_rejects_truncated_stream() {
+        // The streaming reader this replaced returned the bytes it had and no
+        // error, so a truncated chunk read back short.
+        let data = noisy_bytes(100_000);
+        let compressed = deflate_compress(&data, 6).unwrap();
+        for cut in [compressed.len() - 1, compressed.len() / 2, 3] {
+            assert!(
+                deflate_decompress(&compressed[..cut], data.len()).is_err(),
+                "truncated to {cut} of {} bytes",
+                compressed.len()
+            );
+        }
+    }
+
+    #[test]
+    #[cfg(feature = "deflate")]
+    fn deflate_compress_roundtrips_incompressible_data() {
+        // Random-looking input compresses to slightly more than it started
+        // as; the output must still fit the pre-sized buffer (or grow).
+        let mut x = 0x9E37_79B9_7F4A_7C15u64;
+        let data: Vec<u8> = (0..200_000)
+            .map(|_| {
+                x ^= x << 13;
+                x ^= x >> 7;
+                x ^= x << 17;
+                x as u8
+            })
+            .collect();
+        for level in [0, 1, 6, 9] {
+            let compressed = deflate_compress(&data, level).unwrap();
+            assert_eq!(deflate_decompress(&compressed, data.len()).unwrap(), data);
+        }
+        assert_eq!(
+            deflate_decompress(&deflate_compress(&[], 6).unwrap(), 0).unwrap(),
+            Vec::<u8>::new()
+        );
+    }
+
     #[test]
     #[cfg(feature = "zstd")]
     fn zstd_decompress_rejects_output_exceeding_chunk_size() {
diff --git a/crates/clawhdf5-gpu/Cargo.toml b/crates/clawhdf5-gpu/Cargo.toml
index 3ac02aa..d24aba4 100644
--- a/crates/clawhdf5-gpu/Cargo.toml
+++ b/crates/clawhdf5-gpu/Cargo.toml
@@ -2,6 +2,7 @@
 name = "clawhdf5-gpu"
 version = "2.7.0"
 edition = "2024"
+rust-version.workspace = true
 description = "GPU-accelerated vector operations for rustyhdf5 using wgpu compute shaders"
 license = "MIT"
 repository = "https://git.redclaw.dev/quantumclaw/clawhdf5"
diff --git a/crates/clawhdf5-io/Cargo.toml b/crates/clawhdf5-io/Cargo.toml
index f339c71..53d82f1 100644
--- a/crates/clawhdf5-io/Cargo.toml
+++ b/crates/clawhdf5-io/Cargo.toml
@@ -2,6 +2,7 @@
 name = "clawhdf5-io"
 version = "2.7.0"
 edition = "2024"
+rust-version.workspace = true
 description = "I/O abstraction layer for rustyhdf5"
 license = "MIT"
 repository = "https://git.redclaw.dev/quantumclaw/clawhdf5"
diff --git a/crates/clawhdf5-migrate/Cargo.toml b/crates/clawhdf5-migrate/Cargo.toml
index cf714fd..a518a22 100644
--- a/crates/clawhdf5-migrate/Cargo.toml
+++ b/crates/clawhdf5-migrate/Cargo.toml
@@ -2,6 +2,7 @@
 name = "clawhdf5-migrate"
 version = "2.7.0"
 edition = "2024"
+rust-version.workspace = true
 description = "CLI to migrate SQLite agent memory databases to HDF5 format"
 license = "MIT"
 repository = "https://git.redclaw.dev/quantumclaw/clawhdf5"
diff --git a/crates/clawhdf5-napi/Cargo.toml b/crates/clawhdf5-napi/Cargo.toml
index 28d8bda..c47403f 100644
--- a/crates/clawhdf5-napi/Cargo.toml
+++ b/crates/clawhdf5-napi/Cargo.toml
@@ -2,6 +2,7 @@
 name = "clawhdf5-napi"
 version = "2.7.0"
 edition = "2024"
+rust-version.workspace = true
 description = "Node.js native addon (napi-rs) exposing clawhdf5-agent to TypeScript/JavaScript"
 license = "MIT"
 repository = "https://git.redclaw.dev/quantumclaw/clawhdf5"
diff --git a/crates/clawhdf5-netcdf4/Cargo.toml b/crates/clawhdf5-netcdf4/Cargo.toml
index ad49148..6c38342 100644
--- a/crates/clawhdf5-netcdf4/Cargo.toml
+++ b/crates/clawhdf5-netcdf4/Cargo.toml
@@ -2,6 +2,7 @@
 name = "clawhdf5-netcdf4"
 version = "2.7.0"
 edition = "2024"
+rust-version.workspace = true
 description = "NetCDF-4 read support built on rustyhdf5 — pure Rust, no C dependencies"
 license = "MIT"
 repository = "https://git.redclaw.dev/quantumclaw/clawhdf5"
diff --git a/crates/clawhdf5-py/Cargo.toml b/crates/clawhdf5-py/Cargo.toml
index 8f3ea65..0d54a4f 100644
--- a/crates/clawhdf5-py/Cargo.toml
+++ b/crates/clawhdf5-py/Cargo.toml
@@ -2,6 +2,7 @@
 name = "clawhdf5-py"
 version = "2.7.0"
 edition = "2024"
+rust-version.workspace = true
 description = "Python bindings for rustyhdf5 — a pure-Rust HDF5 library"
 license = "MIT"
 repository = "https://git.redclaw.dev/quantumclaw/clawhdf5"
diff --git a/crates/clawhdf5/Cargo.toml b/crates/clawhdf5/Cargo.toml
index ee5326f..2e2b4b4 100644
--- a/crates/clawhdf5/Cargo.toml
+++ b/crates/clawhdf5/Cargo.toml
@@ -2,6 +2,7 @@
 name = "clawhdf5"
 version = "2.7.0"
 edition = "2024"
+rust-version.workspace = true
 description = "Pure-Rust HDF5 reader/writer — no C dependencies"
 license = "MIT"
 repository = "https://git.redclaw.dev/quantumclaw/clawhdf5"
@@ -30,9 +31,10 @@ name = "parallel_bench"
 harness = false
 
 [features]
-default = ["mmap", "fast-deflate", "provenance"]
+default = ["mmap", "provenance"]
 mmap = ["clawhdf5-io/mmap"]
 parallel = ["clawhdf5-format/parallel", "rayon"]
+# zlib-ng (C, needs cmake) instead of the default pure-Rust zlib-rs.
 fast-deflate = ["clawhdf5-format/fast-deflate"]
 apple-compression = []
 zstd = ["clawhdf5-format/zstd"]
diff --git a/crates/libaec-sys/Cargo.toml b/crates/libaec-sys/Cargo.toml
index 788781a..75a7e8b 100644
--- a/crates/libaec-sys/Cargo.toml
+++ b/crates/libaec-sys/Cargo.toml
@@ -2,6 +2,7 @@
 name = "libaec-sys"
 version = "0.1.0"
 edition = "2024"
+rust-version.workspace = true
 links = "aec"
 
 [build-dependencies]
diff --git a/docs/QUICKSTART.md b/docs/QUICKSTART.md
index c751768..0635ce0 100644
--- a/docs/QUICKSTART.md
+++ b/docs/QUICKSTART.md
@@ -567,4 +567,4 @@ let final_results = confidence::reject_low_confidence(
 
 ---
 
-<p align="center"><em>Built by <a href="https://github.com/redclawsystems">RedClaw Systems</a></em></p>
+<p align="center"><em>Built by <a href="https://git.redclaw.dev/quantumclaw">RedClaw Systems</a></em></p>
diff --git a/docs/USE_CASES.md b/docs/USE_CASES.md
index 1ac85a9..4d0cb4a 100644
--- a/docs/USE_CASES.md
+++ b/docs/USE_CASES.md
@@ -232,4 +232,4 @@ clawhdf5-agent = { version = "2.0", features = ["agent", "float16", "accelerate"
 
 ---
 
-<p align="center"><em>Built by <a href="https://github.com/redclawsystems">RedClaw Systems</a></em></p>
+<p align="center"><em>Built by <a href="https://git.redclaw.dev/quantumclaw">RedClaw Systems</a></em></p>
diff --git a/scripts/ci-test.sh b/scripts/ci-test.sh
index f04dd09..b52c88b 100755
--- a/scripts/ci-test.sh
+++ b/scripts/ci-test.sh
@@ -77,6 +77,50 @@ run_step "cargo clippy (ann parallel)" cargo clippy \
     --features parallel \
     -- -D warnings
 
+# zlib-ng is opt-in (`fast-deflate`; the default is pure-Rust zlib-rs), so
+# nothing above builds it. Keep it compiling and passing.
+run_step "cargo clippy (fast-deflate / zlib-ng)" cargo clippy \
+    -p clawhdf5-format -p clawhdf5-filters -p clawhdf5 \
+    --all-targets \
+    --features clawhdf5-format/fast-deflate,clawhdf5-filters/fast-deflate \
+    -- -D warnings
+
+# The README promises that the core crates build no C by default. Hold it to
+# that: fail if a crate that compiles C (a *-sys crate, cc or cmake) enters the
+# default dependency tree of any of them. clawhdf5-migrate (bundled SQLite),
+# clawhdf5-napi (Node) and clawhdf5-gpu (graphics drivers) are exempt.
+no_c_in_default_build() {
+    local crate found=0
+    for crate in clawhdf5-format clawhdf5-io clawhdf5-filters clawhdf5 \
+        clawhdf5-agent clawhdf5-ann clawhdf5-accel clawhdf5-netcdf4 clawhdf5-cli; do
+        local c_deps
+        c_deps=$(cargo tree -q -p "$crate" -e normal,build --prefix none \
+            | grep -E '^([a-z0-9_-]+-sys|cc|cmake) v' | sort -u)
+        if [ -n "$c_deps" ]; then
+            echo "$crate pulls in C by default:"
+            echo "$c_deps" | sed 's/^/    /'
+            found=1
+        fi
+    done
+    return $found
+}
+run_step "no C in the default build (core crates)" no_c_in_default_build
+
+# The workspace declares a minimum Rust version (rust-version in Cargo.toml);
+# check that it really builds there, so the README badge and the manifests
+# cannot drift from the truth. Separate target dir: a different toolchain
+# would otherwise invalidate the main build.
+msrv_check() {
+    local msrv
+    msrv=$(sed -n 's/^rust-version = "\(.*\)"/\1/p' "$SCRIPT_DIR/../Cargo.toml")
+    [ -n "$msrv" ] || { echo "no rust-version in Cargo.toml"; return 1; }
+    rustup toolchain install "$msrv" --profile minimal >/dev/null || return 1
+    echo "checking with Rust $msrv"
+    CARGO_TARGET_DIR="$SCRIPT_DIR/../target/msrv" cargo "+$msrv" check \
+        --workspace --exclude clawhdf5-py --all-targets
+}
+run_step "MSRV check" msrv_check
+
 # 4. Tests (exclude clawhdf5-py)
 run_step "cargo test" cargo test \
     --workspace \
@@ -90,6 +134,10 @@ run_step "cargo test (ann parallel)" cargo test \
     -p clawhdf5-ann \
     --features parallel
 
+run_step "cargo test (fast-deflate / zlib-ng)" cargo test \
+    -p clawhdf5-format -p clawhdf5-filters -p clawhdf5 \
+    --features clawhdf5-format/fast-deflate,clawhdf5-filters/fast-deflate
+
 # 5. Python interop suites. The h5py writer tests are #[ignore]d so a plain
 #    `cargo test` stays hermetic; run them explicitly here.
 #    On a PEP 668 "externally managed" system h5py can only live in a