tla-proof

tla-proof is a skill for Claude Code, Codex from younes-io/agent-skills. It costs 87 tokens per session (991 once invoked), scanned A, original, MIT.

A workflow for writing and checking TLA+ theorem proofs with TLAPS. TLA+ is a language for describing system behavior, and TLAPS is a tool that checks its proofs.

In plain words
What is it for?
Use it to create or repair theorem and proof blocks, investigate TLAPS failures, and strengthen invariants in TLA+ modules.
Why use it?
It makes clear which proof obligations succeeded, failed, or were skipped, along with the assumptions behind the result.

Skill for Claude CodeCodex

Written for Claude Code and Codex: shipped in a Claude Code plugin, but also agents/openai.yaml present.

Part of the tla-workbenches plugin — 2 skills shipped together

Good fit Use it to create or repair theorem and proof blocks, investigate TLAPS failures, and strengthen invariants in TLA+ modules.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/younes-io/agent-skills/tla-proof
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add younes-io/agent-skills --skill tla-proof
Clone the repo
git clone --depth 1 https://github.com/younes-io/agent-skills

Made for: Claude Code, Codex.

Or install tla-workbenches, the plugin that ships this one along with the rest of its 2 skills.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for tla-proof

README.md
[![agentmods](https://agentmods.dev/badge/skills/younes-io/agent-skills/tla-proof/github.svg)](https://agentmods.dev/skills/younes-io/agent-skills/tla-proof)
Your own site
<a href="https://agentmods.dev/skills/younes-io/agent-skills/tla-proof"><img src="https://agentmods.dev/badge/skills/younes-io/agent-skills/tla-proof/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for tla-proof

Your own site · 80×15
<a href="https://agentmods.dev/skills/younes-io/agent-skills/tla-proof"><img src="https://agentmods.dev/badge/skills/younes-io/agent-skills/tla-proof.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 87 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 991 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00087 $0.00991
Opus 5 $0.00044 $0.00495
Sonnet 5 $0.00017 $0.00198
Haiku 4.5 $0.00009 $0.00099

Measured 11d ago against content hash af795582da26, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-10, from the pricing page.

Security

Grade A, and why

tla-proof scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 11d ago.

The scan reads SKILL.md. This mod also ships 2 executable files (scripts/tests/tlaps_check_test.sh, scripts/tlaps_check.sh), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

plugins/tla-workbenches/skills/tla-proof/SKILL.md · 108 lines

How it starts

The opening of the file, as written. The whole thing — 108 lines — stays where its author put it; the contents beside it link to each section on GitHub.

TLA+ Proof

Outputs

  • Updated proof-bearing TLA+ module(s): *.tla
  • TLAPS run artifacts: .tla-proof/runs/<run-id>/...

Non-Negotiables (Honesty Rules)

  • Never claim full system correctness from a partial proof.
  • Always report what was proved, what failed, and what was omitted.
  • Always surface trust boundaries (ASSUME, AXIOM, omitted proofs, imported facts).
  • Never conflate TLC outcomes with TLAPS outcomes; treat them as different evidence.
  • Always keep theorem statements stable while debugging unless the user approves spec changes.

Workflow (Target Class -> Proof Plan -> TLAPS -> Iterate)

1) Pin Down Target and Trust Boundary

Record:

  • theorem statement(s) in scope
  • proof class: direct fact, inductive invariant, refinement, formula equivalence, safety, liveness
  • assumptions/environment model
  • required imported definitions/lemmas
  • candidate strengthening invariants or helper lemmas if the target does not look inductive yet
  • proof granularity target (quick progress vs fully structured proof)

If theorem intent is ambiguous, state candidate interpretations and choose one explicitly. If the user is refactoring a spec and wants semantic preservation, consider a direct equivalence theorem (F <=> G) instead of only bounded TLC evidence.

2) Draft Minimal Hierarchical Proof Structure

Start with the smallest stable structure:

  • THEOREM ...
  • PROOF
  • SUFFICES, HAVE, CASE, PICK, TAKE, WITNESS, QED as needed

Prefer explicit sub-lemmas over long single-step BY clauses. Use BY DEF ... only for required definitions. Formula-equivalence proofs for refactors are a supported pattern, for example THEOREM F <=> G BY DEF F, G. Match the structure to the proof class:

  • inductive invariant/safety: isolate base case vs step case and split Next by action
  • refinement: state the abstraction relation/refinement mapping and prove init/step obligations separately
  • formula equivalence: start with THEOREM F <=> G; if a one-line proof fails, split into F => G and G => F
  • liveness/starvation freedom: pin down fairness and ranking assumptions before proof search, then expect auxiliary lemmas

Read the full file on GitHub · 108 lines

Files

What ships with it

8 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 11d ago First seen · 108 lines · 87 tokens per session scan A af795582da26

Subscribe to this mod's changes

tla-proof is a skill published in the GitHub repository younes-io/agent-skills (20 stars, last pushed 2mo ago), licensed MIT. It adds 87 tokens to every session and 991 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other skills, from other repositories

automatic-differentiation-advanced

Extend Sounio's automatic differentiation capabilities to higher‑order derivatives, Hessians, differentiation through control flow, and GPU‑accelerated gradient computation, enabling state‑of‑the‑art scientific machine learning.

Sounio-lang/sounio · 50 tokens

epistemic-uncertainty-quantification

Extend Sounio's epistemic types with advanced uncertainty quantification methods, such as confidence intervals, non‑Gaussian distributions, and Dempster‑Shafer evidence combination, positioning the language at the state‑of‑the‑art in measurement science.

Sounio-lang/sounio · 62 tokens

formal-verification-epistemic

Integrate formal verification tools (SMT solvers, Lean) to prove properties of epistemic programs, such as uncertainty bounds, confidence guarantees, and provenance integrity, establishing Sounio as a language with mathematically verified scientific claims.

Sounio-lang/sounio · 55 tokens

lean-axiom-provenance

Find out what a Lean 4 project actually rests on, and why. Reports every theorem reaching a sorry anywhere upstream, everything settled by nativedecide rather than the kernel, and the shortest path from any declaration to any axiom with each hop labelled a statement dependency or a proof dependency. Use when the user…

vince-gonzalez/gonzalgo · 203 tokens

lean-generated-proof-audit

Check whether a machine-generated Lean 4 proof actually proves its theorem. A proof can appear in the environment, pass lake build, and still not have been proved: when elaboration fails Lean admits the declaration carrying sorryAx, which an axiom report cannot tell apart from a sorry somebody typed. Use when auditing…

vince-gonzalez/gonzalgo · 143 tokens

sounio-epistemic-types

Work on Sounio epistemic computing: Knowledge/uncertainty/confidence/provenance, ontology bindings, and dependent epistemic subtyping/proofs; use when editing compiler/src/epistemic/, compiler/src/dependent/, or stdlib/epistemic/.

Sounio-lang/sounio · 71 tokens