Wait for shell bootstrap before dispatching child prompts, serialize provider preprocessing, and make permission callbacks idempotent. Restrict task lists to concrete multistep plans and stop inferring completion from tool activity.
Update regression fixtures and resolve existing lint and test-layout failures. Verified formatting, both presubmit Clippy commands, and 354 targeted nextest tests.
- Gate server requests on available credentials
- Run local child agents directly without a parent run ID
- Include command IDs in Bedrock context and recognize transfer tools
Add the 'Crosscheck Work' experiment to the Agents settings. When enabled,
a reviewer sub-agent is spawned after the main agent finishes a turn (with
no pending tool calls). The reviewer critiques the output using a dedicated
system prompt focused on correctness, simplicity, and code quality. If the
reviewer does not respond with 'LGTM!', its feedback is injected as a
synthetic user query back to the main agent, which must address it. This
loop continues until the reviewer approves or max iterations is reached.
Components:
- Feature flag: CrosscheckWork (enabled in DOGFOOD_FLAGS)
- Settings: agents.experiments.crosscheck_enabled,
agents.experiments.crosscheck_model_id,
agents.experiments.crosscheck_max_iterations
- Settings UI: new 'Experiments' subpage under Agents
- Crosscheck module: app/src/ai/crosscheck/ with prompt, reviewer model
- Controller integration: hooks into AfterStreamFinished when no actions
are queued, triggers reviewer, handles feedback injection
- Provider support: OpenAI-compatible and Bedrock direct invocation
- Safety: max iteration guard, error handling, reset on new user query
rkw6086
Ryan Ward