feat(work): idle-timeout instead of fixed wall-clock timeout for dispatched work

[getappz]

Summary

• agentflare work's single fixed --timeout (default 30 min) killed jobs still making real progress — no way to tell a genuinely stuck agent from one on a long but productive run.
• run_captured/run_headless now take two durations: idle_timeout (kill only when stdout+stderr produce no new bytes for that long — the primary liveness signal) and hard_cap (absolute backstop against a runaway process).
• --timeout becomes the hard cap (default bumped 30min → 6h); new --idle-timeout (default 5 min) is the real signal. A task producing steady output can now run for hours; a stalled one is caught in 5 min instead of the old 30. Failure messages report which one fired.
• agentflare run --print / cli_run_headless pass the same value for both params, preserving their existing flat-deadline behavior exactly (one-off invocations, not dispatched work).

Closes item #20 (agentflare tracker) — interim idle-timeout direction chosen over the larger loopx-style turn/delivery-metadata rearchitecture, which needs restructuring agentflare work around discrete inspectable turns and is a bigger, separate lift.

Test plan

• cargo build -p agentflare — clean
• cargo test -p agentflare --bin agentflare agent_launch:: — 22/22 pass, incl. 2 new tests: output-keeps-arriving survives the idle window; stall-after-real-output is caught by idle, not the hard cap
• cargo test -p agentflare --bin agentflare -- agents:: cli::work:: cli::run:: — 33/33 pass
• cargo clippy -p agentflare --bin agentflare — no new warnings

Summary by CodeRabbit

• New Features

  • Added separate limits for maximum runtime and inactivity, allowing processes that continue producing output to run longer.
  • Added a five-minute idle-output timeout and six-hour maximum runtime default for work tasks.

• Bug Fixes

  • Improved handling of stalled processes by terminating them after inactivity.
  • Timeout messages now indicate whether termination was caused by inactivity or maximum runtime.
  • Results now identify when a process stopped due to idle output.

[coderabbitai]

Note

Reviews paused

It looks like this branch is under active development. To avoid overwhelming you with review comments due to an influx of new commits, CodeRabbit has automatically paused this review. You can configure this behavior by changing the reviews.auto_review.auto_pause_after_reviewed_commits setting.

Use the following commands to manage reviews:

• @coderabbitai resume to resume automatic reviews.
• @coderabbitai review to trigger a single review.

Use the checkboxes below for quick actions:

• ▶️ Resume reviews
• 🔍 Trigger review

📝 Walkthrough

Walkthrough

Headless execution now uses separate hard-cap and idle-output timeouts. Output readers track activity incrementally. Captured results identify idle termination. CLI defaults, call sites, and supervisor limits use the new timeout model.

Changes

Headless timeout handling

Layer / File(s)	Summary
Capture and idle detection src/agent_launch.rs	run_captured drains stdout and stderr in chunks, resets the idle timer when output arrives, enforces the hard cap, and records idle termination. Tests cover sustained output, stalled processes, output preservation, and process-tree termination.
Headless termination reporting src/agent_launch.rs	run_headless accepts separate timeout durations and reports idle termination separately from hard-cap termination. Existing headless tests use the new API.
CLI timeout configuration src/cli/work.rs, src/agents.rs	WorkArgs uses a six-hour hard cap and a five-minute idle timeout. The headless CLI passes one timeout to both controls.
Supervisor timeout alignment src/supervisor.rs	The supervisor job timeout increases to 21,900 seconds and documents the inner work timeout and overhead.

Estimated code review effort: 4 (Complex) | ~45 minutes

Sequence Diagram(s)

sequenceDiagram
  participant run_headless
  participant run_captured
  participant child_process
  participant stdout_stderr_readers
  run_headless->>run_captured: Pass hard cap and idle timeout
  run_captured->>child_process: Start and drain process
  child_process->>stdout_stderr_readers: Produce output chunks
  stdout_stderr_readers->>run_captured: Report activity
  run_captured->>child_process: Kill after idle timeout or hard cap
  run_captured-->>run_headless: Return captured output and termination metadata

Loading

Possibly related PRs

• getappz/agentflare#155: Introduced the headless execution APIs and call sites extended by this change.
• getappz/agentflare#260: Also modifies headless timeout and argument handling for the work workflow.
• getappz/agentflare#398: Also modifies run_headless timeout handling and captured output diagnostics.

Suggested labels: enhancement, rust

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly and concisely describes the main change to use idle timeouts for dispatched work.
Description check	✅ Passed	The description explains the change and includes a detailed test plan, but it omits the optional reviewer notes section.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch task/20

---

_{Comment @coderabbitai help to get the list of available commands.}