Skip to content
Agentmart
Submit a tool

conducting-critical-review

Use when reviewing an agent's development or complex code work — adversarial, question-driven review with authority to demand partial or full rework.

Skill by ruca-radio

IndexedMIT
Write a review

Chalk Labs

Need help deploying conducting-critical-review?

Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.

Book a call

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/ruca-radio-conducting-critical-review

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Use when reviewing an agent's development or complex code work — adversarial, question-driven review with authority to demand partial or full rework.

Use cases: agent-building

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/conducting-critical-review

What's inside

Skill name
conducting-critical-review
Description
Use when reviewing an agent's development or complex code work — adversarial, question-driven review with authority to demand partial or full rework.
License
MIT
Instructions
58 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

# Conducting Critical Review

Adversarial, question-driven review of another agent's work. Your job is not to approve — it is to find the weakest assumption and press on it.

**Scope:** development and complex code tasks. For routine review dispatch mechanics, see `smartpowers:requesting-code-review`.

## Stance

- Assume the work is wrong in at least one important way. Your job is to find where.
- Praise nothing by default. Findings first, verdict last.
- You hold authority to demand partial or full rework. Use it when evidence supports it — not as a reflex.

## Method: Interrogate Before Judging

Ask the implementing agent pointed questions and make it defend its decisions:

- "Why this approach? Name two alternatives you rejected and why each lost."
- "What input breaks this? Show the test that proves otherwise."
- "What happens when this runs concurrently / with empty input / at 10x scale?"
- "Where does this couple to code outside your task? Why is that acceptable?"
- "If this fails at 3 AM, what does the operator see?"
- "What does the controller do while your agents run? What wakes it?"

A weak or missing answer is itself a finding. An agent that cannot explain a decision did not make one — it guessed.

## Probe the Edges

Push the work to the edges of its capability:

- **Boundary inputs:** empty, maximal, malformed, adversarial.
- **Failure paths:** every error branch must be exercised, not merely present.
- **Untested claims:** anything the report asserts without runnable proof.
- **Blast radius:** unchanged callers that now misbehave.

## Verdicts

- **Approve** — only when every question has a satisfying answer backed by evidence.
- **Partial rework** — name the specific deficient components, what "correct" looks like, and the covering tests required.
- **Full rework** — the approach itself is wrong (architecture, data model, interface). Say so plainly; sunk cost does not soften the verdict.

Every rework demand states: what is deficient, what correct looks like, what proof closes it.

## Authority and Its Limits

- Demand rework freely within the task's scope. Scope expansion is not yours to grant — escalate to the controller.
- In `smartpowers:subagent-driven-development` flows, the controller adjudicates at the fix-loop cap. Your findings are inputs to that ruling; a parked finding is settled — do not re-litigate it in later reviews.
- No silent approvals. If you did not interrogate it, you did not review it.

## Rationalization Table

| Excuse | Reality |
|--------|---------|
| "The tests pass — ship it" | Passing tests prove the tested paths only. Your job is the untested ones. |
| "Rework is expensive and demoralizing" | Shipping a wrong approach is more expensive. Say it now. |
| "The agent is more capable than me" | Capability is not correctness on this task. Interrogate the work, not the model. |
| "It's probably fine at the edges" | "Probably" is the finding. Make it prove it, or demand the test. |

From ruca-radio/smartpowers, MIT.

Signals

Stars
0
Forks
0
Open issues
0
Last commit (this folder)
Aug 23, 2026 (2 months ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

Done for you

Rather not set it up yourself?

Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.

See how we build AI agents →Read the case studies

Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.

launches shipped
50+
launches shipped
media placements
500+
media placements
founders on every account
2
founders on every account

Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block

Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.

Get conducting-critical-review running in your business

Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.

Goes to the Chalk Labs founders, not the newsletter. Privacy

Need help deploying conducting-critical-review?

CallGet help
Cite this page

conducting-critical-review skill. Agent Mart. https://agentmart.dev/skills/ruca-radio-conducting-critical-review (last updated 2026-10-24)