Implements GitHub Issues based on existing execution plans.
Install
mkdir -p .claude/skills/coder && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/16537" && unzip -o skill.zip -d .claude/skills/coder && rm skill.zipInstalls to .claude/skills/coder
Activation
This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.
Implement a planned Pipeline GitHub Issue by number. Use when Codex is asked to execute an active plan end-to-end, update code/tests/docs, mark plan steps complete, run the relevant verification commands, and report results without editing issue labels, closing the Issue, moving the plan, or committing.Key capabilities
- →Implement a planned Pipeline GitHub Issue by number
- →Update code, tests, and documentation
- →Mark plan steps complete
- →Run relevant verification commands
- →Report results without editing issue labels or committing
How it works
The skill reads a GitHub Issue and its active execution plan, then follows the plan steps to implement the issue, update tests and docs, and run verification commands.
Inputs & outputs
When to use coder
- →Implementing planned GitHub issue
- →Updating tests and docs for issues
- →Running verification scripts
About this skill
Coder
Use this skill when the user (or the manager subagent) asks to implement a planned GitHub Issue.
The user must provide an Issue number as argument. If none is provided, ask for one.
Required Context
Read these before taking action:
AGENTS.md- The Issue itself:
gh issue view <number> -c— body, comments, labels. - The active execution plan:
docs/exec-plans/active/issue-<number>-*.md. - The
issueskill:.codex/skills/issue/SKILL.md— for label/lifecycle conventions.
Read additional docs only when needed for the task.
Read the relevant docs/product-specs/ pages before changing or testing product behavior.
Reading the Issue
gh issue view <number> -c
gh issue view <number> --json title,body,labels,assignees,url
Treat the body as the authoritative "what" and the execution plan as the authoritative "how".
Workflow
- Read the Issue with
gh issue view <number> -c. - Read the active execution plan in
docs/exec-plans/active/(look forissue-<number>-*.md). - If the plan is missing and the caller has not indicated the trivial-frontend or docs flow, stop and tell the caller — the planner phase has not completed.
- No-plan modes: for
Flow: trivial-frontendorFlow: docs, work directly from the Issue body, comments, and parent epic. Do not invent an exec plan. Still satisfy the relevant lint/build/test gate below. - Implement the Issue end-to-end:
- Write the code changes described in the plan (or, in trivial-frontend mode, derived from the Issue).
- Add or update automated tests where applicable.
- After Rust changes:
cargo clippy --all -- -D warningsmust pass. - After TypeScript changes:
npx tsx scripts/lint-docs.tsmust pass. - For frontend changes: run the relevant frontend build (check
package.jsonscripts or the affectedpackages/<pkg>/README) and confirm it succeeds. - Run
$test-fastand fix all failures before reporting done.
- For implementation work attached to an epic, add a user-stories doc at
docs/user-stories/epic-<epic>/<issue>-<slug>.mdand link it fromdocs/user-stories/index.mdin the same PR (ISSUE_PROTOCOL §6). Update an existing story for a user-visible bug fix. Docs-only work does not require a story. - Mark each step in the exec plan as completed as you go (edit the plan file). No-plan modes have no plan to mark.
- Update affected documentation (product specs, design docs, generated reference) per the plan's "Docs to Update" section. In no-plan modes, update docs when behavior actually changed.
- Report to the caller what was done, including test results.
Rules
- Do not edit issue labels. The manager owns lifecycle transitions.
- Do not assign or close the Issue.
- Do not commit. The manager will commit the implementation together with the label change.
- Do not ask the user for approval during implementation. Execute the plan.
- Do not move the execution plan to
docs/exec-plans/completed/— the manager does that during Phase 4. - Follow
AGENTS.md. Never commit tomain. Always work on the feature branch already opened for this Issue. - Respect dependency order. If the Issue depends on unfinished work, explain and stop.
- If the Issue body or plan references a Figma URL, use it as an implementation reference. Epic QA verifies the rendered page later.
- When implementation changes behavior, update the relevant product spec or design doc.
- Log unrelated bugs found during implementation in
docs/exec-plans/known-bugs.md— do not fix them inline. - Log structural shortcuts in
docs/exec-plans/tech-debt-tracker.md— do not fix them inline.
Execution Plan Adherence
Follow the execution plan closely:
- Implement each step in the order specified.
- If the plan includes a test strategy, follow it.
- If you encounter something the plan did not anticipate, use your best judgment and note the deviation in your report and as a comment on the Issue (
gh issue comment <number> --body "<note>").
Output
When done, report:
- Issue number and title
- What was implemented
- Tests added or updated
- Test results (pass/fail) including
cargo clippyand$test-fastoutcomes - Docs updated
- Any deviations from the plan (also noted on the Issue)
- Any blockers or concerns
When not to use it
- →When the task involves editing issue labels
- →When the task involves committing or pushing changes
- →When the task involves moving the plan to completed
Limitations
- →Cannot edit Issue labels
- →Cannot commit or push changes
- →Cannot move the plan to completed
How it compares
This workflow automates the execution of a pre-defined plan for a GitHub Issue, unlike manual implementation that requires direct human oversight for each step.
Compared to similar skills
coder side by side with the closest alternatives in the catalog.
| Skill | Installs | Updated | Safety | Difficulty |
|---|---|---|---|---|
| coder (this skill) | 0 | 4mo | Review | Intermediate |
| agent-implementer-sparc-coder | 1 | 7mo | Review | Intermediate |
| scaffold-feature | 0 | 5mo | Review | Intermediate |
| deepwiki-rs | 25 | 11mo | Review | Intermediate |
Try saying
Example prompts that trigger this skill in your AI assistant.
You might also like
agent-implementer-sparc-coder
ruvnet
Agent skill for implementer-sparc-coder - invoke with $agent-implementer-sparc-coder
scaffold-feature
niklasbrandt
Scaffold a complete full-stack feature: FastAPI endpoint, dashboard Web Component, i18n keys, test stubs, and documentation checks.
deepwiki-rs
sopaco
AI-powered Rust documentation generation engine for comprehensive codebase analysis, C4 architecture diagrams, and automated technical documentation. Use when Claude needs to analyze source code, understand software architecture, generate technical specs, or create professional documentation from any programming language.
rust-errors
EpicenterHQ
Rust to TypeScript error handling patterns for Tauri apps. Use when defining Rust errors that will be passed to TypeScript, handling Tauri command errors, or creating discriminated union error types.
lint
Azure
>-
code-review
jonatron55
Instructions for reviewing changes and ensuring quality before completion. Use when asking for a review or before committing changes.