Index of every page on this site: llms.txt
All skills
expo avatar

/llp-review

@f428a37
by expoexpo/exact4.7k stars
132
Run it yourself

Run the LLP 0005 review loop for an LLP document — orchestrate reviews from other model families, write one provenance-tracked artifact per review actually received under llp/reviews/, and never fabricate a review or accept on the author's behalf. Review intensity scales with stakes; the author decides when it's enough.

  • 1 file
  • 2.9 KB
  • Updated 2 weeks ago
  • GitHub
Use this Skill: https://skilld.dev/gh/expo/exact/llp-review

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

SKILL.md

≈84 tokens for metadata: the name and description. ≈653 when used: this file.

Description uses 4.2% of example budget

Before choosing a Skill, your Agent reads its name and description. All available Skills share that space.

  • A shorter description leaves more room for other Skills. This entry exceeds our 1% size suggestion.

In our Claude Code example, all Skill names and descriptions share 8,000 characters. This Skill uses ≈334 characters, or 4.2%.

The 1% threshold is a size suggestion. Longer descriptions can still fit.

Your model, settings, and other Skills decide how much text your Agent can read.

Example settings and source

The example uses a 200k-token context and default Claude Code settings. The count includes the name, description, separators, and when_to_use when present. Codex also counts local file paths.

Skit's source and limits: Codex 0.160.1, Claude Code 2.1.292.

llp-review

<!-- @ref https://github.com/ccheever/llp/blob/v0.5.2/llp/0000-linked-literate-programming.explainer.md#review-honesty — the rules this enforces --> <!-- @ref https://github.com/ccheever/llp/blob/v0.5.2/llp/0005-rfc-process.guide.md#review-artifacts — artifact + provenance header format -->

Review an LLP the way LLP 0005 prescribes: as much critical review as the stakes warrant, captured honestly. Different model families catch different blind spots, so a single session cannot stand in for a multi-family loop — this skill makes the honest path the only path.

Invoke as /llp-review <number | slug | filename>. Bare: list Draft/Review LLPs and ask.

Trigger

A document enters Review (the author's opt-in to a formal loop), or review is explicitly requested. Otherwise offer review scaled to stakes (LLP 0005 §How much review to buy) — don't impose it.

Invariants

  • MUST NOT fabricate a review, record one that wasn't actually provided, or mark a loop complete by intent.
  • MUST NOT count the session that authored the draft as an independent reviewer — if this session wrote the document, say so and gather reviews elsewhere.
  • MUST NOT change **Status:** or accept/reject — propose transitions; the author decides.
  • MUST write one artifact per review actually received: llp/reviews/NNNN-slug.<family>.md, opening with the LLP 0005 provenance header (family · provider/runtime · date · redacted? · method). Re-reviews by the same family get a dated section, never an overwrite.
  • MUST NOT send content to an external provider without an explicit human action; redact secrets by default.

Workflow

Recipe (advisory) — locate and read the document fully. If this session can act as one independent family reviewer (and didn't author the draft), write that review using the LLP 0005 standard prompt. For each further input the author wants — pasted manual review, sub-agent on a different family, CLI runner — prepare the prompt and stop until it is actually provided. Structure each review body: overall assessment · strengths (cite sections) · concerns (with severity and what would resolve each) · suggestions · open questions · recommended next step.

Artifact

Review files under llp/reviews/ with provenance headers, plus a summary of the feedback and a proposed next step (revise and stay Draft / gather another family / ready for the author to accept).

Hand-offs

  • Feedback that amounts to a new design direction → a fresh LLP via llp-create.
  • The author applies status changes; this skill never does.

Source: SKILL.md on GitHub

No rule matched.

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

No third-party reports yet.

Signed by skilld at f428a37. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub yesterday.

Activeupdated 2 weeks ago
source
ccheever/llp@v0.5.2

README badge

README badge for expo/exact/llp-review