Files
clawmates/crates/cm-runtime/tests/shell_exec.rs
T
Omar SobhandClaude Opus 4.8 3554a3aaf2
ci / gates (push) Successful in 5s
ci / frontend (push) Successful in 23s
ci / rust (push) Failing after 27s
ci / e2e (push) Has been skipped
CI: remove k8s stages, fix the Docker-level pipeline green
Survey + fixes so the pipeline passes at the Docker level (no k8s).

- Remove k8s: drop the `sandbox-k8s` job (kind/Calico/--features k8s-tests) and the
  "Helm chart lints" gate step. release.yml was already k8s-clean.
- Rust job:
  - `cargo fmt --all` — fix pre-existing formatting drift (fmt --check was failing).
  - clippy -D warnings: fix 3 lib warnings (cm-brain sort_by_key→Reverse, cm-api
    fleet.rs doc list indentation, node_rules map_or→is_none_or).
  - Regenerate the .sqlx offline cache (was missing the cm-runtime run_loop test
    query → offline compile failed). DB-backed tests use testcontainers at runtime.
  - Set SQLX_OFFLINE=true on the rust + e2e jobs so query! macros compile against
    the committed cache deterministically (no DB needed at compile time).
- Frontend job:
  - Fix the 1 ESLint error (useAgentTelemetry: no setState-synchronously-in-effect;
    tag the slice with agentId + derive null on mismatch).
  - Fix 2 stale panel-params tests (`terminal` is a valid app id now; assert the
    current APP_IDS + use a genuinely-unknown id for the reject case).

Verified locally: fmt clean, clippy --all-targets -D warnings clean (offline),
frontend lint 0 errors, tsc clean, 86/86 frontend tests pass, build OK.

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
2026-06-26 18:15:31 -07:00

233 lines
7.1 KiB
Rust

//! Agents execute code in their REAL hardened sandbox: shell.exec runs in
//! a per-agent container (Docker driver), output returns to the model as a
//! step result, and /home/agent state persists across calls in a run.
use std::process::Command;
use std::sync::Arc;
use cm_domain::{
AccessPolicy, Agent, AgentId, AgentStatus, Role, RunState, User, UserId, Workspace, WorkspaceId,
};
use cm_llm::ScriptedProvider;
use cm_runtime::{Runtime, RuntimeConfig, SandboxManager};
use cm_sandbox::DockerDriver;
const IMAGE: &str = "clawmates/agent-base:dev";
const SCENARIOS: &str = r#"
[[scenario]]
marker = "[[scenario:shell]]"
[[scenario.turns]]
events = [
{ type = "tool_use", name = "shell.exec", input = { command = "id -u && echo tick > /home/agent/state" } },
]
[[scenario.turns]]
events = [
{ type = "tool_use", name = "shell.exec", input = { command = "cat /home/agent/state" } },
]
[[scenario.turns]]
events = [
{ type = "text", text = "Ran both commands." },
]
"#;
fn ensure_image() {
let exists = Command::new("docker")
.args(["image", "inspect", IMAGE])
.output()
.expect("docker available")
.status
.success();
if !exists {
let root = env!("CARGO_MANIFEST_DIR");
let status = Command::new("docker")
.args([
"build",
"-t",
IMAGE,
"-f",
&format!("{root}/../../images/agent-base/Dockerfile"),
&format!("{root}/../../images/agent-base"),
])
.status()
.expect("docker build runs");
assert!(status.success(), "agent-base image build failed");
}
}
#[tokio::test]
async fn shell_exec_runs_in_the_agent_sandbox_with_persistent_home() {
ensure_image();
let pool = cm_testkit::test_pool().await;
let ws = Workspace {
id: WorkspaceId::new(),
name: "Acme".into(),
plan: "team".into(),
};
cm_db::repo::workspaces::insert(&pool, &ws).await.unwrap();
let owner = User {
id: UserId::new(),
workspace_id: ws.id,
email: format!("{}@acme.test", UserId::new()),
role: Role::Owner,
display_name: "Owner".into(),
created_at: time::OffsetDateTime::UNIX_EPOCH,
};
cm_db::repo::users::insert(&pool, &owner).await.unwrap();
let agent = Agent {
id: AgentId::new(),
workspace_id: ws.id,
name: "Scout".into(),
job_title: "Analyst".into(),
system_prompt: String::new(),
avatar: String::new(),
accent: String::new(),
wallpaper: String::new(),
managed_by: owner.id,
status: AgentStatus::Online,
};
cm_db::repo::agents::insert(&pool, &agent, &AccessPolicy::default())
.await
.unwrap();
let driver = DockerDriver::connect().expect("docker reachable");
let sandboxes = Arc::new(SandboxManager::new(
Arc::new(driver),
pool.clone(),
"local",
IMAGE,
));
let rt = Runtime::new(
pool.clone(),
Arc::new(ScriptedProvider::from_toml(SCENARIOS).unwrap()),
RuntimeConfig::basic("scripted", 1024).with_sandboxes(sandboxes.clone()),
);
let session = cm_db::repo::sessions::create(&pool, agent.id, ws.id, "Shell")
.await
.unwrap();
let started = rt
.send_message(session.id, "run it [[scenario:shell]]")
.await
.unwrap();
let mut rx = started.events;
while let Ok(envelope) = rx.recv().await {
if matches!(
envelope.event,
cm_runtime::RunEventBody::RunCompleted { .. } | cm_runtime::RunEventBody::Error { .. }
) {
break;
}
}
let run = cm_db::repo::runs::get(&pool, started.run_id).await.unwrap();
assert_eq!(run.state, RunState::Completed);
// Step outputs prove kernel-level facts: uid 10001 inside, and the
// second call saw the first call's file — same sandbox, same run.
let outputs: Vec<serde_json::Value> = sqlx::query_scalar(
"SELECT s.output FROM steps s
JOIN messages m ON m.id = s.message_id
WHERE m.session_id = $1 AND s.tool_name = 'shell.exec'
ORDER BY s.seq",
)
.bind(session.id.as_uuid())
.fetch_all(&pool)
.await
.unwrap();
assert_eq!(outputs.len(), 2);
assert_eq!(outputs[0]["exit_code"], 0, "first: {}", outputs[0]);
assert!(
outputs[0]["stdout"].as_str().unwrap().contains("10001"),
"sandbox runs as uid 10001: {}",
outputs[0]
);
assert_eq!(outputs[1]["exit_code"], 0, "second: {}", outputs[1]);
assert!(
outputs[1]["stdout"].as_str().unwrap().contains("tick"),
"home persists across execs: {}",
outputs[1]
);
sandboxes.shutdown().await;
}
#[tokio::test]
async fn shell_exec_without_a_sandbox_runtime_reports_a_tool_error() {
let pool = cm_testkit::test_pool().await;
let ws = Workspace {
id: WorkspaceId::new(),
name: "Acme".into(),
plan: "team".into(),
};
cm_db::repo::workspaces::insert(&pool, &ws).await.unwrap();
let owner = User {
id: UserId::new(),
workspace_id: ws.id,
email: format!("{}@acme.test", UserId::new()),
role: Role::Owner,
display_name: "Owner".into(),
created_at: time::OffsetDateTime::UNIX_EPOCH,
};
cm_db::repo::users::insert(&pool, &owner).await.unwrap();
let agent = Agent {
id: AgentId::new(),
workspace_id: ws.id,
name: "Scout".into(),
job_title: "Analyst".into(),
system_prompt: String::new(),
avatar: String::new(),
accent: String::new(),
wallpaper: String::new(),
managed_by: owner.id,
status: AgentStatus::Online,
};
cm_db::repo::agents::insert(&pool, &agent, &AccessPolicy::default())
.await
.unwrap();
let rt = Runtime::new(
pool.clone(),
Arc::new(ScriptedProvider::from_toml(SCENARIOS).unwrap()),
RuntimeConfig::basic("scripted", 1024),
);
let session = cm_db::repo::sessions::create(&pool, agent.id, ws.id, "NoBox")
.await
.unwrap();
let started = rt
.send_message(session.id, "run it [[scenario:shell]]")
.await
.unwrap();
let mut rx = started.events;
while let Ok(envelope) = rx.recv().await {
if matches!(
envelope.event,
cm_runtime::RunEventBody::RunCompleted { .. } | cm_runtime::RunEventBody::Error { .. }
) {
break;
}
}
// The run survives: the tool reports its error to the model, which
// finishes the scenario.
let run = cm_db::repo::runs::get(&pool, started.run_id).await.unwrap();
assert_eq!(run.state, RunState::Completed);
let statuses: Vec<String> = sqlx::query_scalar(
"SELECT s.status FROM steps s
JOIN messages m ON m.id = s.message_id
WHERE m.session_id = $1 AND s.tool_name = 'shell.exec'
ORDER BY s.seq",
)
.bind(session.id.as_uuid())
.fetch_all(&pool)
.await
.unwrap();
assert!(!statuses.is_empty());
assert!(
statuses.iter().all(|s| s == "error"),
"steps must record the failure: {statuses:?}"
);
}