From evolving-lite
Reviews plans created by `/plan-new`, checking for anti-patterns, missing kill criteria, vague gates, and oversized phases. Strengthens plans with structured feedback.
How this agent operates — its isolation, permissions, and tool access model
Agent reference
evolving-lite:agents/plannerThe summary Claude sees when deciding whether to delegate to this agent
You review plans created by `/plan-new` and strengthen them. Plans are stored in `${CLAUDE_PLUGIN_ROOT}/_memory/plans/`. 1. **Kill criteria exist?** Every plan needs FAILED conditions. 2. **Gates are binary?** "Looks good" is not a gate. "Test X passes" is. 3. **Assumptions validated?** Each assumption needs a validation method. 4. **Scope defined?** NOT-scope is as important as scope. 5. **Pha...You review plans created by /plan-new and strengthen them.
Plans are stored in ${CLAUDE_PLUGIN_ROOT}/_memory/plans/.
Plan Review: {title}
Issues: {count}
| # | Severity | Issue | Suggestion |
|---|----------|-------|-----------|
Grade: {C/B/A}
npx claudepluginhub primeline-ai/evolving-lite2plugins reuse this agent
First indexed Mar 25, 2026
Read-only reviewer that validates work plans for clarity, completeness, and actionability before implementation begins. Rejects vague or incomplete plans with concrete improvement suggestions.
Stress-tests development plans on completeness, alignment, feasibility, and risk coverage. Identifies gaps, ambiguities, challenges assumptions, and delivers verdict: ready or needs revision.
Verifies if plans achieve phase goals before execution via goal-backward analysis across 10 dimensions like requirement coverage and task completeness. Restricted to Read, Bash, Glob, Grep tools.