lean-ai-formalization
Skill3 starsChanged 24 days agoApache-2.0View on GitHub
USE FOR: Formal verification of AI systems: agentic AI safety, alignment, high-stakes AI, evolving agents. DO NOT USE FOR: AI methodology itself (@lean-ai-formalization). TRIGGERS: AI formalization, agentic safety, alignment proof.
In this repository65 other files it gives its agents.
- AGENTS.mdAGENTS.md
- ai-guard-judge-evalSkill
- applied-data-information-securitySkill
- applied-engineering-disciplinesSkill
- applied-legal-reasoningSkill
- epistemic-discovery-engineSkill
- epistemic-mappingSkill
- hol-metamath-adapterSkill
- isabelle-isar-adapterSkill
- itp-cross-portingSkill
- lean-applied-reasoningSkill
- lean-axiom-closureSkill
- lean-bigraph-modelingSkill
- lean-bisectSkill
- lean-blueprintSkill
- lean-buildSkill
- lean-causal-reasoningSkill
- lean-competitive-mathSkill
- lean-convergence-verificationSkill
- lean-core-onlySkill
- lean-dag-visualizerSkill
- lean-deep-auditSkill
- lean-doc-feedbackSkill
- lean-doc-improvementSkill
- lean-doc-requirementsSkill
- lean-enforcementSkill
- lean-formalization-breakdownSkill
- lean-gatewaySkill
- lean-integration-protocolSkill
- lean-knowledge-formalizationSkill
- lean-math-analysisSkill
- lean-math-discreteSkill
- lean-math-dynamicalSkill
- lean-math-foundationsSkill
- lean-math-optimizationSkill
- lean-math-stochasticSkill
- lean-mweSkill
- lean-nested-learningSkill
- lean-package-researchSkill
- lean-pedagogical-expositionSkill
- lean-proof-refactorSkill
- lean-proof-repairSkill
- lean-proof-reviewSkill
- lean-proofSkill
- lean-prSkill
- lean-quality-engineSkill
- lean-replay-verificationSkill
- lean-reportSkill
- lean-researchSkill
- lean-retroactive-auditSkill
- lean-retro-methodologySkill
- lean-review-councilSkill
- lean-security-formalizationSkill
- lean-setupSkill
- lean-specificationSkill
- lean-tautology-triageSkill
- lean-upstream-portingSkill
- lean-vendor-substrateSkill
- lean-zettelkastenSkill
- math-product-managementSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 0 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 49
STOP if the property has not been formally stated — delegate to @lean-specification first.
- Very specific: line 50
STOP if Mathlib does not yet have the needed primitives — escalate to @lean-research.
- Very specific: line 51
STOP if the AI system is agentic/emergent — delegate dynamics modelling to @lean-ai-formalization first.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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.
Files
Lean 4 AI Systems Formalization
Guide to formally verifying properties of AI systems, from safety envelopes to alignment guarantees.
Routing
- USE FOR: Formal verification of AI systems — agentic AI safety, alignment, high-stakes AI, evolving agents, neural network properties, and AI governance constraints. Use when formalizing safety envelopes, trust dynamics, multi-agent composition, AI act compliance, reward specifications, alignment properties, or any AI system property that must be formally verified.
- DO NOT USE FOR: AI methodology itself (use @lean-ai-formalization or @lean-ai-formalization); security-specific formalisation (use @lean-security-formalization); knowledge formalisation (use @lean-knowledge-formalization).
- TRIGGERS: AI formalization, agentic safety, alignment proof, high-stakes AI, neural network property, AI governance.
Workflow
- Identify the AI property to formalise: safety, alignment, agentic invariant, or neural-network property.
- Pick the matching section of the body (verification framework, formal-property catalog, alignment-axiom set).
- Produce the Lean statement + proof skeleton; verify against the current Mathlib pin before delegating to
@lean-proof. - Hand off: to
@lean-proof-reviewonce the proof compiles, to@lean-enforcementfor the CI gate, to@lean-zettelkastenfor the pattern.
Recovery & STOP
- STOP if the property has not been formally stated — delegate to
@lean-specificationfirst. - STOP if Mathlib does not yet have the needed primitives — escalate to
@lean-research. - STOP if the AI system is agentic/emergent — delegate dynamics modelling to
@lean-ai-formalizationfirst.
Handoffs
- Predecessors:
agent:gateway,skill:lean-research. - Successors:
skill:lean-proof-review,skill:lean-enforcement,skill:lean-zettelkasten,skill:lean-doc-feedback.
Detailed reference
Full content for lean-ai-formalization lives in
references/lean-ai-formalization-handbook.md.
Load that file when the skill is convened; the SKILL.md only carries
the dispatch contract and the parts index.
| Section | Topic |
| Part 1 | AI Safety Properties |
| Part 2 | Agentic AI |
| Part 4 | High-Stakes AI Verification |
| Part 5 | Neural Network Properties |
| Part 6 | Multi-Agent Composition |
| Part 7 | Commonsense and Causal Reasoning |
| Part 8 | Research Council Integration |
See also
../../references/lean-ai-formalization-handbook.md— Full handbook (extracted from this skill)../lean-proof-review/SKILL.md— Successor../lean-enforcement/SKILL.md— Successor../lean-zettelkasten/SKILL.md— Successor../lean-doc-feedback/SKILL.md— Successor
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
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.