Generate and post an implementation plan for a GitHub issue with a plan-pending label.
npx skills add https://github.com/rjskene/pipeline --skill plan-issue-rjskene
Please help me install this Agent Skill. Skill: plan-issue Source: https://github.com/rjskene/pipeline/tree/main/skills/plan-issue Command: npx skills add https://github.com/rjskene/pipeline --skill plan-issue-rjskene
This Skill converts a GitHub issue into a structured, ready-to-execute implementation plan while enforcing trust boundaries and preventing untrusted automation.
authorAssociation and refuses auto-planning for untrusted openers, surfacing the issue for human triage instead.recommended_path from trusted classification feedback).plan-pending label.needs-browser issues, requires machine-checkable predicates the browser-evaluation sub-skill can run.Tell the AI to run /pipeline:plan-issue <issue_number> to produce and post an implementation plan comment for that issue and mark it as plan-pending.
High-intent search queries and answers about installing and using this skill.
To generate an implementation plan from a GitHub issue, trigger the plan-issue pipeline command with the issue number. It drafts a canonical plan comment, publishes it, and applies a plan-pending label to the issue.
Yes, automated issue planning can be restricted to trusted contributors using a trusted-opener gate. It checks the author association and refuses auto-planning for untrusted openers, surfacing them for human triage.
Label-driven workflow automation selects a planning path by reading specific GitHub issue labels. It can also use a cached recommended path from prior trusted classification feedback to determine the correct planning mode.
To update an existing implementation plan, the workflow detects trusted feedback on an earlier plan and rewrites it. The revised comment includes a required Changes from previous plan section to ensure revision-safe updates.
GitHub issues requiring browser testing enforce label-aware browser predicates during implementation planning. This requires machine-checkable predicates that a browser-evaluation sub-skill can execute to validate the conditions.
After posting an implementation plan, the workflow handles the terminal state by applying a plan-pending label to the GitHub issue. This communicates that the issue is awaiting execution of the published plan.
tomzxScore and rank open GitHub issues using the RICE framework.
mrtolkienDispatch fresh subagents to execute independent plan tasks with two-stage reviews.
agentsopeEnforce explicit counters and exit predicates to terminate LM-driven agent loops.
rjskeneClassify GitHub issues and apply pipeline triage labels.