agentleFS
on-keydayon-keyday/agent-harness/.agents/skills/independent-review/SKILL.md

independent-review

Skill2 starsChanged 2 days agoMITView on GitHub

Use when you want a second opinion that does not inherit your framing — adversarial review of a diff or design, catching plausible-but-wrong work, or questioning whether the whole approach is right. Explains when a built-in subagent suffices vs. when to spawn a truly independent agent. For the harness-cli commands themselves, see the harness-cli skill.

14 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use security-review.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • SentrySkill

    security-review

    Finds exploitable application security vulnerabilities in code changes. Use for Warden security scans, appsec review, OWASP-style checks, authentication or authorization bugs, injection, XSS, SSRF, path traversal, secrets, unsafe crypto, webhook verification, open redirects, or sensitive data exposure.

    Also in 4 projects that use this

The file

We split this skill into the rules it gives an agent

1 rule. 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 10 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Ask for findings and evidence only — the critic must not decide for you.

    line 34

“Common” is measured: kept by at least 6 in 10 of the 10 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

independent-review (adversarial second opinion)

A reviewer who shares your framing shares your blind spots. The value of a second opinion is independence — it reasons from the work itself, not from your justification of it. Reach for this when you need to catch plausible-but-wrong output, or to surface that the premise itself is wrong, not just the details.

Pick the right independence level

NeedUseWhy
Mechanical checks — untested claims, boundary breaks, spec violations, contradictionsClaude Code built-in subagent (@-mention it, or name it in your prompt)Fresh context window, zero setup. The parent crafts the subagent's task prompt, so it inherits your framing — but that is fine here, because mechanical flaws don't depend on independence.
Questioning the premise — "is this approach even right?", subtle design-sense callsA harness-spawned agent (harness-cli session new -d)Boots a fresh claude on its own worktree with no shared context — true independence. A subagent cannot give this: its prompt is written by the very agent whose work is under review, so it inherits that agent's blind spots and self-justification.

The built-in subagent is claude-only. The harness-spawned path works for any agent and is the one to use when independence is load-bearing.

The adversarial brief

Whichever you pick, tell the critic to attack, in two layers:

  1. Refute the details — untested claims, boundary breaks, spec violations, contradictions. Most defects live here, and a subagent handles them well.
  2. Refute the premise — list at least 3 ways the whole approach could be wrong, plus any alternative framing. Tell it explicitly not to assume the approach is correct.

Ask for findings and evidence only — the critic must not decide for you. A summary substitutes for your judgment and can mislead (LLM summaries go wrong plausibly); pointers to real flaws do not. Give the critic the artifact, not your reasoning — reasoning from scratch is the whole point.

Independence has a limit worth stating: a critic is itself an LLM and shares some of the worker's blind spots, so it reliably raises the floor on mechanical flaws but does not replace your judgment on the subtle, premise-level calls.

Driving a harness-spawned critic

Spawn it, hand it the artifact (push the files, or point it at a committed branch — never your justification), send the adversarial brief, then end your turn — its findings arrive through the inbox hook, not a blocking wait.

For the actual commands — session new -d, agent send, file push, the chat.<short-id> inbound-topic convention, and the async/inbox rules — see the harness-cli skill (harness-cli skill harness-cli, or .claude/skills/harness-cli/SKILL.md). This skill is the when / why; harness-cli is the how.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.