Repository navigation
fix(skills): plan-collab replies to every ask and covers stop and tasks - #81
Merged
Merged
Conversation
Dogfooding the local seat, an Ask answered with only a suggestion stayed "Queued" on the page: the page settles a prompt only on agent.reply. - Every ask.request and agent.request ends with agent.reply; document the `intent` field (chat answers in the reply, edit suggests then replies) and the no-selection insert case. - Cover run.stop and plan.tasks.request/plan.tasks.result, which the page sends and the skill did not mention. - Note that ghost completions need `--model @alias` in practice: the next keystroke cancels a request before a hand-driven agent answers. Regenerated plugin copies with tools/plugin-gen. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Problem and intent
I dogfooded
archdev plan collabagainst archdev.ai on 2026-10-21, with an agent following this skill:The page (firstlanding
services/go/archdev/web/src/plan-collab/plan-session.ts) marks a prompt done only when it receivesagent.reply {request: <id>}. The skill told agents to answerask.requestwith only a suggestion, so an agent that followed it exactly left every Ask queued forever.The page also sends two signals the skill never mentions:
run.stop {run}(Stop agent) andplan.tasks.request {request, revision}(Break into tasks). An agent following the skill ignored both.What changed
ask.requestdocuments itsintentfield.chatmeans answer in the reply without editing;editmeans suggest, then reply. Everyask.requestandagent.requestnow ends withagent.reply, with an example command.rangeandquote) inserts at that position.run.stop(sendrun.statusstopped) andplan.tasks.request→plan.tasks.result {request, revision, tasks[]}, using the page'sDraftTaskshape (key, title, risk, area, dependsOn).--model @aliasin practice. In one session the page sent 56complete.requestand 28 cancels, and every hand-drivencompletefailed with "canceled, replaced or unknown".tools/plugin-gen(npm run build).npm run checkreports that the 148 generated files match the catalog.Scope: skill text only. Risk: low. User impact: agents using the skill clear the page's queued prompts and handle Stop and Break into tasks.
Testing
npm run checkpasses.npm testreported one failing assertion locally. I couldn't rerun it to identify which, so this PR'spluginsCI run is the gate. Main is green.🤖 Generated with Claude Code