skills.rest
davepoon
davepoonCOMMUNITY
Explore all skills by this author →
Website𝕏@davepoonGitHub

coding-agent-pm

Plan, delegate, and review coding-agent implementation work with scoped briefs and acceptance criteria.

3.4k|487|Updated Jul 25, 2025
Search Tags:#acceptance criteria#task delegation#Validation#scope control#coding agent#Code review#Claude code
One-click install
npx skills add https://github.com/davepoon/buildwithclaude --skill coding-agent-pm
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: coding-agent-pm
Source: https://github.com/davepoon/buildwithclaude/tree/main/plugins/all-skills/skills/coding-agent-pm
Command: npx skills add https://github.com/davepoon/buildwithclaude --skill coding-agent-pm

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Delegating implementation work to coding agents like Claude Code, Codex, or OpenClaw often produces scope drift, unvalidated changes, and diffs that are hard to trust. This Skill turns vague requests into bounded agent tasks with explicit ownership, acceptance criteria, and validation requirements.

Core Features & Use Cases

  • Agent Brief Template: Defines the shipped outcome, owned and excluded files, forbidden actions, project context, and acceptance criteria before work starts.
  • Review Checklist: Reviews agent diffs in priority order covering regressions, missing criteria, weak validation, scope drift, and security risks.
  • Recovery Prompt: Stops a drifting agent and forces it to map every change back to the original acceptance criteria before further edits.
  • Use Case: When asking an agent to add OAuth login, produce a brief that owns only src/auth and login routes, forbids edits to billing or migrations, and requires tests for success, denied OAuth, and linked accounts.

Quick Start

Ask the assistant to turn your implementation request into a bounded coding-agent brief with scope, acceptance criteria, and validation steps.

Frequently Asked Questions about coding-agent-pm

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I write a good task brief for a coding agent?▼

State the shipped outcome in one sentence, list owned and excluded files, forbid destructive actions, and write checkable acceptance criteria. Require the agent's final response to include changed files, validation results, and remaining risks.

How do I review code generated by Claude Code or Codex?▼

Review the diff in priority order: behavioral bugs or regressions first, then missing acceptance criteria, weak validation, scope drift, and security risks. Reject unrelated refactors and require every change to map back to the original criteria.

What should I do when a coding agent drifts off task?▼

Pause implementation and ask the agent to map every changed file back to the original acceptance criteria, marking each change as required, optional, or unrelated. Do not allow further edits until that map is complete.

Does this work with coding agents other than Claude Code?▼

Yes, the brief template, review checklist, and recovery prompt apply to any coding agent including Codex and OpenClaw. The structure is agent-agnostic and focuses on scope control and validation rather than tool-specific commands.

When should I not delegate a task to a coding agent?▼

Avoid delegating when you cannot define bounded file ownership or checkable acceptance criteria, or when the task touches shared risky modules without a clear do-not-edit list. Vague, unbounded requests produce untrustworthy diffs.

Related Skills

Explore More Software Engineering Skills→
michael-watsonmichael-watson
8

rpi-research

Analyze codebase architecture, dependencies, and constraints into a research document.

Community
Intermediate
indykishindykish
2

review

Analyze branch diffs against origin/main for merge safety issues.

Community
Advanced
Euda1mon1aEuda1mon1a

automated-code-fixer

Detect and correct code defects across test suites, linters, type checkers, and security scanners.

Community
Advanced
Handy369Handy369

subagent-guide

Determine when to split complex requests into parallel subagent workstreams.

Community
Intermediate