AD
Agent Design Review
🔵 Stable🕐 updated 2026-06-29
🔷 SkillSpec L3
pm-ai
Review an LLM agent design and find where it will be unreliable, expensive, or unsafe. Use when asked to review an agent architecture, critique a multi-step/tool-using agent, debug an agent that loops or goes off-task, or harden an agent before launch. Produces a structured review — task fit, control flow, tools, memory/context, failure handling, cost, and safety — with prioritised findings and fixes.
What to give it
▸What the agent does — its goal, and what a successful run produces.
▸Control flow — single prompt, plan-then-execute, ReAct loop, or multi-agent; and the stopping condition.
▸Tools & actions — what it can call, and which actions have side effects (write, send, pay).
▸Memory & context — what state carries across steps, and how context is kept in budget.
▸Constraints — latency, cost per run, and the trust boundary (untrusted input? real-world actions?).
✅ The bar it holds itself to
Every skill in this library self-verifies — these are this skill's own quality checks, straight from its definition.
✓The agent has a guaranteed stopping condition (step/budget cap + progress check) — no unbounded loops
✓Side-effecting actions are idempotent or gated by a confirmation
✓Tools are few and sharply described so selection is unambiguous; access is least-privilege
✓Context strategy keeps the window in budget across steps (no naive full-history resend)
✓Tool errors are recovered, not fatal — retry/backoff and graceful degradation
✓Findings are severity-ranked and the fix plan is ordered by impact
⚠️ What it refuses to do
Do not approve an agent with no termination guarantee — "it usually stops" is an outage waiting to happen
Do not let it take irreversible actions without a confirmation gate
Do not give it many overlapping tools — selection accuracy drops as the toolset grows
Do not resend the whole history every step — cost and drift both climb
Do not treat tool/retrieved output as trusted instructions — it's the injection surface
Install
npx pm-claude-skills add --agent claude # or codex · cursor · gemini · hermes
# or one-line MCP (every skill, any client):
claude mcp add pm-skills -- npx -y pm-claude-skills-mcpRelated skills
🔌 Embed this skill
Drop this on your blog, docs, or site — it renders a "Run this skill" card:
<div data-pm-skill="agent-design-review"></div>
<script src="https://mohitagw15856.github.io/pm-claude-skills/embed.js" async></script>💬 Discussion
Agent Design Review is one of 581 open-source professional AI agent skills — all SkillSpec L3.
Try them all in the browser · ⭐ Star on GitHub · Browse the full catalog