P1 backend: chat persistence, tc-llm providers, runtime loop, gateway SSE

- tc-db: sessions/messages/steps/runs/run_events repos (atomic seq assignment,
  history with ordered step traces, journal replay-from-offset); migration 0003
- tc-llm: provider-neutral ChatRequest/LlmEvent; ScriptedProvider (scenario
  TOML, word-level deltas, multi-turn tool legs — ships in production for
  e2e/air-gap smoke), AnthropicProvider (Messages SSE), OpenAiCompatProvider
  (vLLM/Ollama/llama.cpp); opt-in live tests via TC_LIVE_LLM=1
- tc-runtime: run loop with persist-before-emit event journal, real built-in
  clock.now tool, step rows on the reply message, tool-error resilience,
  broadcast channels for live attach
- tc-api: agent CRUD + settings/full (tenant-isolated, RBAC'd, audited),
  sessions create/list/history?tools=true, POST /api/gateway SSE with
  monotonic ids and exact resumeFrom journal replay (tested equal to live)
- teamclaw-server: config-driven provider factory

83 Rust tests green, all against real Postgres / real TCP.

Co-Authored-By: Claude Fable 5 <[email protected]>

crates/tc-llm/tests/live.rs

@@ -0,0 +1,71 @@
+//! Opt-in tests against real inference endpoints. Activated with
+//! `TC_LIVE_LLM=1` plus `ANTHROPIC_API_KEY` (Anthropic) or
+//! `TC_OPENAI_COMPAT_URL` (e.g. a local Ollama at
+//! `http://127.0.0.1:11434/v1` with `TC_OPENAI_COMPAT_MODEL` set).
+//! CI runs these in a dedicated credentialed job; the default suite uses
+//! the scripted provider, which exercises the identical seam.
+
+use futures::StreamExt;
+use tc_llm::{
+    AnthropicProvider, ChatMessage, ChatRequest, ChatRole, ContentPart, LlmEvent, LlmProvider,
+    OpenAiCompatProvider,
+};
+
+fn live_enabled() -> bool {
+    std::env::var("TC_LIVE_LLM").as_deref() == Ok("1")
+}
+
+fn simple_request(model: &str) -> ChatRequest {
+    ChatRequest {
+        system: "Answer in exactly one short sentence.".into(),
+        messages: vec![ChatMessage {
+            role: ChatRole::User,
+            parts: vec![ContentPart::Text("Say the word 'pong'.".into())],
+        }],
+        tools: vec![],
+        model: model.into(),
+        max_tokens: 64,
+    }
+}
+
+async fn collect_text(provider: &dyn LlmProvider, request: ChatRequest) -> String {
+    let mut stream = provider.stream(request).await.expect("stream opens");
+    let mut text = String::new();
+    while let Some(event) = stream.next().await {
+        if let LlmEvent::TextDelta(t) = event.expect("stream event") {
+            text.push_str(&t);
+        }
+    }
+    text
+}
+
+#[tokio::test]
+async fn anthropic_streams_text() {
+    if !live_enabled() {
+        eprintln!("skipped: set TC_LIVE_LLM=1 to run");
+        return;
+    }
+    let Ok(key) = std::env::var("ANTHROPIC_API_KEY") else {
+        eprintln!("skipped: ANTHROPIC_API_KEY not set");
+        return;
+    };
+    let provider = AnthropicProvider::new(key);
+    let text = collect_text(&provider, simple_request("claude-haiku-4-5-20251001")).await;
+    assert!(text.to_lowercase().contains("pong"), "got: {text}");
+}
+
+#[tokio::test]
+async fn openai_compat_streams_text() {
+    if !live_enabled() {
+        eprintln!("skipped: set TC_LIVE_LLM=1 to run");
+        return;
+    }
+    let Ok(url) = std::env::var("TC_OPENAI_COMPAT_URL") else {
+        eprintln!("skipped: TC_OPENAI_COMPAT_URL not set");
+        return;
+    };
+    let model = std::env::var("TC_OPENAI_COMPAT_MODEL").unwrap_or_else(|_| "qwen2.5:0.5b".into());
+    let provider = OpenAiCompatProvider::new(url, None);
+    let text = collect_text(&provider, simple_request(&model)).await;
+    assert!(text.to_lowercase().contains("pong"), "got: {text}");
+}