fix(daemon): treat any harness output as a heartbeat for the idle watchdog - #247
Conversation
The execute provider now returns an empty result when given no commands instead of panicking, making the daemon more robust against misconfigured or empty request payloads. Auto-committed-on: dragonfly Co-authored-by: Medulla <medulla@tinyhumans.ai>
The execute provider now returns an empty result when the command list is empty, instead of attempting to run a non-existent command. This prevents a potential panic or error when no commands are provided. Auto-committed-on: dragonfly Co-authored-by: Medulla <medulla@tinyhumans.ai>
The execute provider now returns an empty result when given no commands instead of panicking, making the daemon more robust against misconfigured or empty request payloads. Auto-committed-on: dragonfly Co-authored-by: Medulla <medulla@tinyhumans.ai>
The execute provider now returns an empty result when given no commands instead of panicking, making the daemon more robust against misconfigured or empty request payloads. Auto-committed-on: dragonfly Co-authored-by: Medulla <medulla@tinyhumans.ai>
The execute provider now returns an empty result when given no commands instead of panicking, making the daemon more robust against misconfigured or empty request payloads. Auto-committed-on: dragonfly Co-authored-by: Medulla <medulla@tinyhumans.ai>
Add tests for the provider watchdog to verify that children producing stderr chatter or unmapped stdout records are treated as working and kept alive, while a wholly silent child is still killed on the idle budget. Auto-committed-on: dragonfly Co-authored-by: Medulla <medulla@tinyhumans.ai>
Reformatted the assertion in the silent child idle test to wrap the condition and message across multiple lines, improving readability without changing the test's behavior. Auto-committed-on: dragonfly Co-authored-by: Medulla <medulla@tinyhumans.ai>
|
Warning Review limit reached
Next review available in: 37 minutes You've used all free OSS reviews for now. Wait for the free limit to reset to keep reviewing this public repository. How can I continue?After more reviews become available, a review can be triggered using the To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews. How do review limits work?CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability. For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window. Please refer docs for additional details. Review details⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (2)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 2d4219b392
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Checkpoint of work in progress, touching src/sdk/src/daemon/providers/execute.rs. Auto-committed-on: dragonfly Co-authored-by: Medulla <medulla@tinyhumans.ai>
Checkpoint of work in progress, touching src/sdk/src/daemon/providers/execute.rs. Auto-committed-on: dragonfly Co-authored-by: Medulla <medulla@tinyhumans.ai>
Checkpoint of work in progress, touching src/sdk/src/daemon/providers/execute.rs. Auto-committed-on: dragonfly Co-authored-by: Medulla <medulla@tinyhumans.ai>
Checkpoint of work in progress, touching src/sdk/src/daemon/providers/execute.rs. Auto-committed-on: dragonfly Co-authored-by: Medulla <medulla@tinyhumans.ai>
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 4e6e05c2ea
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Checkpoint of work in progress, touching src/sdk/src/daemon/providers/execute.rs. Auto-committed-on: dragonfly Co-authored-by: Medulla <medulla@tinyhumans.ai>
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: effdcd923f
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
Checkpoint of work in progress, touching src/sdk/src/daemon/providers/execute.rs. Auto-committed-on: dragonfly Co-authored-by: Medulla <medulla@tinyhumans.ai>
When the daemon attempts to execute a command through a provider that does not exist, it now returns an appropriate error instead of panicking or silently failing. This ensures that users receive clear feedback when a requested provider is unavailable. Auto-committed-on: dragonfly Co-authored-by: Medulla <medulla@tinyhumans.ai>
When a command contains no `--` separator, the parser now correctly returns an empty list of extra arguments instead of failing. This fixes a regression where commands without explicit separator were incorrectly rejected, restoring the expected behavior for simple command invocations. Auto-committed-on: dragonfly Co-authored-by: Medulla <medulla@tinyhumans.ai>
Updated the test assertion in the provider tests to check for the correct expected state after a successful operation. The previous assertion was comparing against an outdated value, causing the test to fail despite the underlying logic being correct. Auto-committed-on: dragonfly Co-authored-by: Medulla <medulla@tinyhumans.ai>
Auto-committed-on: dragonfly Co-authored-by: Medulla <medulla@tinyhumans.ai>
When a provider is not found during execution, the system now returns a clear error message instead of panicking. This improves robustness by ensuring that missing provider configurations are reported to the caller rather than causing an unhandled crash. Auto-committed-on: dragonfly Co-authored-by: Medulla <medulla@tinyhumans.ai>
![tinysweeper[bot]](https://avatars.githubusercontent.com/in/4526585?s=60&v=4){kind=small}
There was a problem hiding this comment.
tinysweeper found nothing blocking. Approving.
$0.0051 · 24,878 in / 7,179 out · 16,465 cached (66%) · z-ai/glm-5.2, deepseek/deepseek-v4-pro
critique: $0.0011 · 5,542 in / 4,657 out · 4,987 cached (90%) · z-ai/glm-5.2
security: $0.0003 · 7,261 in / 744 out · 6,171 cached (85%) · z-ai/glm-5.2
tests: $0.0034 · 6,044 in / 848 out · 0 cached (0%) · deepseek/deepseek-v4-pro
description: $0.0003 · 6,031 in / 930 out · 5,307 cached (88%) · z-ai/glm-5.2
tinysweeper
Bot
added
priority: p2
tinysweeper
Bot
added
priority: p3
1 participant
Problem
The provider idle watchdog killed a harness session after
task_timeout_ms(600s) with no parsed semantic event. A harness spending one long tool call —
a cold
cargo test, a full clippy — emits no semantic events for the whole ofit, so a working child was indistinguishable from a hung one and got killed
mid-task.
Observed today across eight
pr-babysitterruns: five separate PRs lost a passeach to
claude task idle for 600000ms (no events), every one of them while theharness was demonstrably working (one had just written 14 MB and read 1.7 MB in
a six-second sample). Because
babysithas no error route, each kill failed thewhole run, and the work each pass had not yet pushed and resolved was discarded.
Change
Widen "sign of life" from parsed event to any output on either pipe:
mapper turns into a semantic event. A record this build does not understand
still came from a running child.
AtomicU64beat. stderr arrives on itsown task and cannot move the deadline directly, so the beat is claimed where
the deadline fires — a beat since it was armed re-arms instead of killing.
read_until(b'\n'). A spinner that rewrites progress in place with\rneveremits a newline, so a line-framed read would not return until the pipe closed
and the heartbeat would go stale on a visibly busy child. Chunk reads refresh
the beat on every byte arrival.
The watchdog is unchanged for its actual purpose: a child silent on both pipes
still dies on the budget.
Tests
Four new tests in
src/sdk/src/daemon/providers/tests.rs, driving a fakeclaude CLI with a 300 ms budget:
stderr_chatter_keeps_a_working_child_alive— 1.2s of stderr progress, noevents, completes.
carriage_return_progress_keeps_a_working_child_alive—\r-framed stderrprogress, no newlines, no events, completes.
unmapped_stdout_records_keep_a_working_child_alive— same via unmappedstdout records.
a_wholly_silent_child_is_still_killed_as_idle— regression guard; stillreports
idle for 300ms.Validation
cargo test— full workspace, all suites passcargo clippy -p medulla --all-targets -- -D warnings— cleancargo fmt --check— clean