agentleFS
clarknova99clarknova99/multi-provider-harness/.agents/skills/idea-review/SKILL.md

idea-review

Skill1 starsChanged 4 months agoMITView on GitHub

Run a multi-persona adversarial review of a proposed idea before it earns a spec — a "red team" that tries to kill the idea, so only ideas that survive scrutiny enter a build cycle. Use when evaluating whether a risky or expensive idea is worth building, or pressure-testing a proposal after generative brainstorming. Optional, not a required gate.

3 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

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

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    When you run this gate, it produces a docs/DECISIONS.md entry, kill or proceed — that's the link between the idea and the decision log.

    line 27
  2. Distinctive:

    Don't let the panel become a formality.

    line 28

“Common” is measured: kept by at least 6 in 10 of the 3 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

Skill: Idea review (adversarial pre-spec gate)

The purpose of this gate is to find the reason an idea should NOT be built before any time goes into a spec or code. This is the structured "red team" step: it produces a decision, and that decision gets recorded.

This is the adversarial half of the idea workflow, and it is optional. Run brainstorming FIRST if the shape is still unclear (that's the generative half). Then run this gate to attack the resulting design. Brainstorming grows ideas; this gate prunes them. Use it when an idea is risky, expensive, or hard to reverse — skip it for small, obvious changes and go straight to a spec.

How to run it

Convene the relevant personas from .agents/agents/ and give each an explicit adversarial charge. Run them as separate perspectives (in one session, or as parallel subagents if your tool supports it):

  • product-manager — Does this serve a real user need, or is it a solution looking for a problem? Does it fight the product's core differentiator? What's the smaller version?
  • principal-engineer — What does this cost architecturally? Does it force a parallel code path or violate an invariant? What breaks at scale?
  • ux-design-reviewer — Will users discover and understand it? Does it add chrome or confusion?
  • qa-analyst — How would this fail in the field? What's expensive to verify?

Each persona's job is to find the strongest reason to NOT build it. Steelman the objections; don't rubber-stamp.

The gate

After the panel:

  • Killed — record why in docs/DECISIONS.md (a rejected idea is a decision worth remembering, so it isn't re-proposed). Stop.
  • Survives — record the surviving rationale AND the serious objections raised (with how they'll be mitigated) in docs/DECISIONS.md, then write the spec in docs/specs/ with Status: DRAFT.

Conventions to keep

  • When you run this gate, it produces a docs/DECISIONS.md entry, kill or proceed — that's the link between the idea and the decision log.
  • Don't let the panel become a formality. If nothing is challenged, the review didn't happen.

Output

A short verdict: KILLED (with reason) or PROCEED (with surviving rationale, top objections, and mitigations), plus the docs/DECISIONS.md entry and — if proceeding — the DRAFT spec path.

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.