agentleFS
winchxyzwinchxyz/idea-to-build/.claude/skills/recheck/SKILL.md

recheck

Skill9 starsChanged 5 months agoMITView on GitHub

Re-critique a project after the build, with real execution results — does the data disconfirm the THESIS or just the tuning? Use when the user types "/recheck" (often with results/data pasted) or asks to re-examine a project in light of how it's actually going.

2 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions2 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 2

    1. dmitribogatenkov/briefproof★ 68same name
    2. zahidakhyar/skills★ 0same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository12 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    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:

    The results are whatever the user provided: after /recheck (as a slash command, $ARGUMENTS) — pasted data, metrics, or a description of how …

    line 14
  2. Distinctive:

    Default to suspecting the thesis, not the knobs.

    line 16
  3. Distinctive:

    If the verdict is pivot or kill, point back to the rejected alternatives in the context file as the place to look — don't let the user …

    line 21
  4. Distinctive:

    If there is no prior brainstorm/context for the project, say so and offer to run a fresh critique instead.

    line 28

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “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

/recheck — re-critique with execution results (Phase 5, re-critique mode)

The user built (or started building) a project from an earlier idea-to-build brainstorm and now has real results. They want the critic to re-examine it in light of that data — not to tune parameters, but to question whether the core thesis still holds. This is the loop that catches calibration-myopia while it's happening, which a one-time pre-build critique cannot.

Steps

  1. Read core/CLAUDE.md (coordinator role) and core/agents/critic.md — especially its "Re-critique mode (invoked with results)" section.
  2. Load the project's memory: read context/<slug>.md for the chosen idea, its core thesis, the original critique, and the rejected alternatives. If it's unclear which project this is, ask. If the built project has a FRAME-LOG.md (written by /scaffold's frame-check guardrail), read it too — it's the run-record of in-build tuning-vs-thesis calls and their predictions. Use it as evidence: a long trail of altitude:in-build tuning entries with no thesis-test is itself a calibration-myopia signal.
  3. The results are whatever the user provided after /recheck (as a slash command, $ARGUMENTS) — pasted data, metrics, or a description of how it's going. If none is given, ask what actually happened.
  4. Become the critic yourself, in-context (do not spawn a sub-agent, do not soften), and run re-critique mode under one overriding question — does the data disconfirm the THESIS, or only the TUNING?:
    • Default to suspecting the thesis, not the knobs. Make the user prove the premise is still alive before touching a parameter.
    • Outcome vs. proxy: are the numbers (good or bad) the real outcome or a proxy that can move while the real thing doesn't?
    • Build fidelity: does what was built still implement the chosen thesis, or did it drift into a generic version that was never the plan?
    • Small-N as an excuse: "not enough data to conclude" is fair for the parameters, but must not defer the structural question (is the thesis even coherent / tested correctly?).
    • Verdict: 🟢 keep (thesis holding) / 🟡 pivot the thesis (premise looks wrong — change approach, don't tune) / 🔴 kill (thesis disconfirmed).
  5. If the verdict is pivot or kill, point back to the rejected alternatives in the context file as the place to look — don't let the user tunnel a dead approach.
  6. Append the re-critique and verdict to context/<slug>.md (dated, in English), so the project's log carries it.

Rules

  • This is a judgment step, in-context, exactly like Phase 5 — the same critic, a new occasion. Don't tune parameters for the user; make them prove the thesis is still alive first.
  • Chat in the user's language; write the context-file entry in English (§6).
  • If there is no prior brainstorm/context for the project, say so and offer to run a fresh critique instead.

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.