Commit Graph
7 Commits
Author SHA1 Message Date
can1357 b48aeacf25 fix(coding-agent/eval): made cells own every bridge call they start
- Split the Python bridge signal: the raw abort reaches tools (so
  subagents die with the turn) while a shielded signal governs how long
  the host waits, so a cancel can no longer settle a cell on top of a
  still-running isolation merge.
- Mirrored the contract in the JS runtime: abort in-flight tool calls
  immediately, then drain any deferExternalAbort phase before killing
  the worker, and refuse new bridge calls once cancelled.
- Held a finished worker result until the run's tool calls drain. A
  floated agent() previously settled the cell at once, dropping the
  run's abort listener and leaving the subagent running with nothing
  able to cancel it.
- Added regression coverage for all three, each verified to fail
  without its fix.
2026-08-03 19:55:20 +02:00
can1357 c7a3010d7c fix(coding-agent/eval): ensured tool bridge receives unshielded abort signal
- Prevent eval agent bridge calls from inheriting the kernel abort shield.
- Unset maxRuntimeMs in runEvalAgent to properly inherit task.maxRuntimeMs.
2026-08-03 18:57:22 +02:00
can1357 ccc9b900cf fix(eval): forked subagent kernel resets away from shared sessions
- Subagents inherit the parent's eval session id, so a child's
  reset: true destroyed the co-owned kernel and every sibling's
  interpreter state mid-session.
- resolveOwnerScopedSessionKey now routes a reset from a non-exclusive
  owner onto a deterministic per-owner fork key: the requester gets a
  fresh private kernel, co-owners keep the shared one, and the fork
  stays sticky for that owner until its teardown reaps it.
- Applied across Python, JavaScript, Ruby, and Julia executors; JS
  contexts gained an owner registry plus disposeVmContextsByOwner,
  wired into EvalRunner.disposeKernels and SDK session teardown.
- Covered by pure key-resolution contracts and an end-to-end JS test:
  co-owner reset forks, shared state survives, fork is sticky, and
  per-owner dispose reaps only the fork.
2026-07-31 20:13:23 +02:00
roboomp 31098f9248 fix(tools): closed spilled output descriptors on error/abort paths
OutputSink.dump() was the only path that closed the spill Bun.FileSink.
The bash and Python executors re-throw on failure and their finally
blocks never closed the sink, so any large-output command that errored
leaked the artifact descriptor until an unrelated read (e.g. a SKILL.md
load) hit EMFILE.

Added an idempotent OutputSink.dispose() that closes the sink exactly
once (awaiting any in-flight sink creation, guarding post-finalize
resurrection) and wired it into every executor's finally block.

Fixes #6463
2026-07-24 03:44:12 +00:00
can1357 7a2b027e47 fix(eval): keep completion aborts interruptible 2026-07-14 18:45:03 +02:00
roboomp a86c1ec46d fix(eval): shield agent bridge aborts
- Deferred eval cancellation while Python bridge calls are paused so already-started agent() subagents can finish and persist output.
- Rejected new Python bridge calls after an external abort is pending to prevent post-abort fan-out waves.
- Added regression coverage for the bridge shield and Python parallel agent() interruption path.

Fixes #5005
2026-07-10 01:21:39 +00:00
can1357 a94cadf723 refactor(coding-agent): standardized evaluation kernel and executor architecture
- Extracted common kernel and executor logic into `BaseKernel` and `executor-base` to eliminate duplicated implementations for Julia, Python, and Ruby.
- Migrated shared operational workflows--including session namespacing, environment filtering, and result mapping--to centralized backend helpers.
- Consolidated runtime discovery and resolution logic into a unified `runtime-env` utility module.
- Simplified language-specific modules by delegating subprocess lifecycle, IPC, and configuration management to the newly established base classes.
2026-06-22 07:22:10 +02:00