Repository navigation
feat(agent-rules): stall-watching rule + AgentWorkforce orchestration overlay (Phase 3) - #238
Open
AgentRelayBot wants to merge 2 commits into
Open
AgentRelayBot wants to merge 2 commits into
AgentRelayBot wants to merge 2 commits into
Conversation
…ration overlay Phase 3 of docs/specs/grok-bot-dynamic-prompt.md, stacked on Phase 2. Rules 2026.11.1. - sections/stall-watch.md (grok-bot, lead; base, decision D7): keep a list of delegated work; follow up on agents silent for 30 minutes, or 10 minutes after a spawn with no ack; route blockers; tell the person who owns the work without waiting to be asked; check even on runs with no new messages. One follow-up per agent per interval, one summary note per run. - overlays/agentworkforce-grok-bot.md: the reviewed AgentWorkforce grok-bot overlay (D7). Grok bots do no coding and no real decisions; all coding goes to Claude and Codex agents; they only spawn, route and relay; problems are surfaced to Khaliq. Not part of the base bundle: it takes effect only when an organization admin PUTs it, which on production needs Khaliq's OK. - No automatic heartbeat routine (D6, conservative): docs note that quiet periods need an opt-in scheduled routine. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
|
Important Review skippedAuto reviews are disabled on base/target branches other than the default branch. Please check the settings in the CodeRabbit UI or the
|
Carries the reworked Phase 1 and Phase 2 (main merged in, review findings addressed, private references removed) into Phase 3; no history rewrite. Rules 2026.11.1 and its pin are unchanged. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Phase 3 of
docs/specs/grok-bot-dynamic-prompt.md. Stacked on #237, which is stacked on #236. Rules version2026.11.1.Rules
agent-rules/sections/stall-watch.md, base, audiencesgrok-botandlead). A bot keeps a list of the work it has delegated. It follows up on agents that have not reported: 30 minutes since the last message, or 10 minutes after a spawn with no acknowledgement. It routes blockers, and it tells the person who owns the work, without waiting to be asked, when an agent is gone, failed or silent after one follow-up, or when work is blocked on a decision, credential, approval or outage. It also runs the check on runs with no new messages. At most one follow-up per agent per interval, and one summary note per run.agent-rules/overlays/agentworkforce-grok-bot.md). Grok bots do no coding work and make no real decisions; all coding goes to Claude and Codex coding agents; they only spawn, route and relay. Problems go to Khaliq (khaliq-gant).Decisions applied
docs/agent-rules.md). Applying it in production needs Khaliq's OK, so I did not apply it.Verification
npx tsc --noEmit: pass.test/agent-rules.test.ts: 7 tests pass, including the new Phase 3 test. It checks that stall-watch reachesgrok-botandleadbut notworker, that the overlay text passes the lint and composes under "Organization rules" after stall-watch, and that the pin, version and CHANGELOG checks hold for 2026.11.1.test/mcp.test.tsand both Grok test files: 79 tests pass. This PR has no code or migration changes.No merge, deploy or flag flip without Khaliq's direct OK.
🤖 Generated with Claude Code
Open in [code]smith · Need help on this PR? Tag
@codesmith-botwith what you need.Summary by cubic
Adds Phase 3 of the
grok-botdynamic prompt spec: a stall-watching rule so bots follow up on delegated work that stalls, plus a reviewed orchestration-only overlay for AgentWorkforce. Rules version bumps to2026.11.1; there are no code or migration changes.New rules
stall-watch(base, audiencesgrok-botandlead): bots track delegated work, follow up on agents silent for 30 minutes or unacknowledged 10 minutes after a spawn, route blockers, and proactively tell the work owner when an agent is gone, failed, or still silent after one follow-up, or when work is blocked on a decision, credential, approval, or outage. Runs with no new messages still do the check; at most one follow-up per agent per interval and one summary note per run.overlays/agentworkforce-grok-bot.md: Grok bots do no coding and make no real decisions; all coding goes to Claude or Codex coding agents, and problems surface to Khaliq. It is not in the base bundle; it takes effect only when an AgentWorkforce owner or admin PUTs it (command indocs/agent-rules.md), which on production needs Khaliq's direct OK.Decisions applied
Written for commit c88450e. Summary will update on new commits.