agentsclimarketplace
Sign in

Lean blueprint

Skill r-irbe/proof-skills/skills/lean-blueprint

USE FOR: generating Lean blueprint, annotating theorems with @[blueprint], scaffolding blueprint directory, building blueprint LaTeX, rendering blueprint web. DO NOT USE FOR: writing Lean proofs (use @lean-proof), opening Lean/Mathlib PRs (use @lean-pr). TRIGGERS: blueprint, leanblueprint, LeanArchitect, dependency-graph.From its SKILL.md

Install
npx -y skills add r-irbe/proof-skills --skill lean-blueprint

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
  • runs commandsInstructs the agent to run 3 commands, including `lake build :blueprint` and 2 more.

SKILL.md

3.8 KB, 875 tokens by cl100k_base, as published. Nobody here has run it

lean-blueprint

⚠️ MANDATORY (hot-tier): the gates in §Behavioural rules and the Persist step in §Workflow are CI-enforced. Skipping Persist = incomplete, regardless of artefact quality (FSIA-R-11-09).

Routing

  • USE FOR: generating a navigable formal blueprint from a Lean 4 codebase; annotating definitions and theorems with @[blueprint]; scaffolding blueprint/ with leanblueprint new; building blueprint LaTeX via lake build :blueprint; rendering the dependency graph and web view.
  • DO NOT USE FOR: authoring new Lean proofs (use @lean-proof); opening Lean/Mathlib PRs (use @lean-pr); editing prose docs (use @lean-doc-improvement).
  • TRIGGERS: blueprint, leanblueprint, LeanArchitect, dependency-graph.

Behavioural rules (G-*)

  • G-1 (MUST): The blueprint pipeline MUST run the five stages in order (analyze → annotate → scaffold → extract → render). [Trace: AC-01]
  • G-2 (MUST): Every @[blueprint] annotation MUST carry a matching LaTeX statement in content.tex. [Trace: AC-02]
  • G-3 (MUST NOT): The skill MUST NOT mutate files outside blueprint/ and the annotated .lean modules. [Trace: AC-03]
  • G-4 (SHOULD): The skill SHOULD fan out per-module annotators when ≥6 modules are in scope. [Trace: AC-04]
  • G-5 (MUST): lake build :blueprint MUST succeed before render. [Trace: AC-05]
  • G-6 (MUST): leanblueprint pdf and leanblueprint web MUST both produce artefacts under blueprint/. [Trace: AC-06]
  • G-7 (SHOULD NOT): The skill SHOULD NOT re-annotate modules already carrying complete @[blueprint] coverage. [Trace: AC-07]
  • G-8 (MUST): On any guard failure the skill MUST escalate per §Recovery & STOP. [Trace: AC-08]

Workflow

  1. Discover [discover] — read inputs, list Lean modules, locate lakefile.lean and prior blueprint/.
  2. Plan [discover] — propose per-module annotator fan-out; STOP if confidence < 80 %.
  3. Annotate [execute] — apply @[blueprint] annotations and LaTeX statements inside owned scope.
  4. Build [execute] — run lake build :blueprint.
  5. Verify [validate] — confirm artefacts exist; max 3 retries then escalate.
  6. Render [execute] — run leanblueprint pdf and leanblueprint web.
  7. Persist [persist] (MANDATORY, FSIA-R-11-09) — commit changes, update state tracker, tick tasks.md. Skipping Persist = incomplete.

Recovery & STOP

  • lake build :blueprint fails ×3 → STOP, escalate to human.
  • Scope drift (touch outside blueprint/ or annotated .lean files) → immediate STOP, re-anchor.
  • Confidence < 80 % on annotation strategy → STOP, ask.
  • Context degradation signals (≥2 from AGENTS.md) → recommend fresh session.

Handoffs

  • Predecessors / successors: see FM handoffs (grammar from ADR-0080).
  • Source spec: specs/lean/blueprint/requirements.md — every G-rule traces to an AC there.
  • Related ADRs: ADR-0076, ADR-0080.

Common failure modes

AI agents commonly: skip the Persist step; over-annotate stable modules; re-run lake build instead of escalating on third failure. Full registry: GUARDRAILS.md §Agent failure taxonomy.

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Gives 0 of the 12 instructions most operations skills give in 875 tokens

Counted across 483 of the 484 authors here whose files we hold, read 2026-08-07

  • Collect monitoring data throughout the simulationin 14 of 483, across 6 files
  • Set the random seed for reproducibilityin 14 of 483, across 6 files
  • Validate simulations against analytical solutionsin 12 of 483, across 4 files
  • Clarify goals, constraints, and inputsin 11 of 483, across 2 files
  • Implement contract tests for integration pointsin 11 of 483, across 2 files
  • Implement strangler fig infrastructure with API gatewayin 11 of 483, across 2 files
  • Audit modernized components for security vulnerabilitiesin 11 of 483, across 2 files
  • Avoid Python blocking calls in processesin 10 of 483, across 3 files
  • Use resource context managers for automatic cleanupin 9 of 483, across 2 files
  • Maintain consistent time unitsin 9 of 483, across 2 files
  • Validate outcomes against success criteriain 8 of 483, across 1 file
  • Analyze the legacy codebase for technical debtin 8 of 483, across 1 file

Said here and by no other author read

  • run pipeline stages in order
  • match annotation with latex statement
  • build blueprint before render
  • produce both pdf and web artifacts
  • escalate on guard failure
  • stop if confidence is low

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.