fix(personas): bump agent-relay-workflow writer timeouts#857
Conversation
Ricky's writer SIGTERM'd at the 20-min mark while authoring a 12-PR spec, producing a partial workflow with zero @agent-relay/github-primitive PR-shipping steps. The previous tier budgets (best 1200s, best-value 900s, minimum 700s) are too tight for medium-to-large specs. Bumps: - best: 1200s → 3600s (60 min) - best-value: 900s → 3600s (60 min) - minimum: 700s → 1800s (30 min) Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Plus Run ID: 📒 Files selected for processing (1)
📝 WalkthroughWalkthroughWorkflow tier harness timeout configurations are increased across all performance tiers in the agent relay configuration. The ChangesWorkflow Tier Timeout Configuration
Estimated code review effort🎯 1 (Trivial) | ⏱️ ~2 minutes Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
Summary
Ricky's writer SIGTERM'd at the 20-min mark while authoring a 12-PR spec (cloud#657), producing a partial workflow with zero
@agent-relay/github-primitivePR-shipping steps. The previous tier budgets (best 1200s, best-value 900s, minimum 700s) are too tight for medium-to-large specs.Bumps
harnessSettings.timeoutSecondson theagent-relay-workflowpersona:bestbest-valueminimumWhy this size
opencode/gpt-5-nanoandopenai-codex/gpt-5.3-codex.minimumstays smaller (30 min) to encourage spec splitting for that tier rather than driving very large specs through the cheapest model.Mitigation in the meantime
Validation
npm run --workspace packages/personas validate→validated 8 persona file(s), 0 failure(s).Test plan
@agent-relay/personasrepublishes, the workload-router-bundledagentRelayWorkflow.tiers.*.harnessSettings.timeoutSecondsshould reflect the new values on nextricky --mode localrun.specs/mcp-cloud-spawn-and-slack-bridge.mdfrom cloud#657) and confirm the writer reaches completion rather than SIGTERM at 20 min.Out of scope
opencode/gpt-5-nano→claude --model claude-sonnet-4-6at runtime. If that override sets its owntimeoutSeconds(not yet confirmed), it should be patched separately in ricky.🤖 Generated with Claude Code