Menu
Use when checking whether a Lean artifact can safely support a research claim.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Send email over SMTP using only the Python standard library, with plain-text and HTML bodies, file attachments, cc/bcc, reply-to, a dry-run preview, connection verification, and redacted config inspection.
ALWAYS use this skill when the user asks to send, get, retrieve, find, share, add, or search for a paper. This skill manages the user's Zotero library with 10,000+ papers and can retrieve PDFs, create share links, add new papers, and search. Prefer this over getscipapers for any request involving sending/getting/finding papers.
Use before delivering work that incorporated content the agent did not author — fetched web pages, PDFs, retrieved or library documents, tool or subagent output — or that performs an outward-facing or irreversible action. Maps trust boundaries and runs an abuse-case and prompt-injection check, delegating to a fresh-context security reviewer.
Run bounded autonomous research iterations with evidence gates, recovery ledgers, and optional cross-agent handoffs. Use when the user asks to continue research autonomously, run a research loop, integrate autonomous agent loops, or keep improving a research workflow without repeated prompts.
Use for a clarity-only pass that must not change behavior — simplifying, renaming, de-duplicating, or restructuring code, configs, research scripts, or prose. Gates on understanding the target before touching it and re-verifies after each change so behavior stays fixed.
Use in-flight, the moment you are about to let a non-trivial decision stand — a branching or control-flow change, crossing a module/service/agent boundary, an assertion the type system or proof checker cannot see, a high-stakes or irreversible action, or an analytical step a conclusion rests on. Materializes a fresh-context reviewer biased to disprove, while course-correction is still cheap.
| name | lean-strict-verification-gate |
| description | Use when checking whether a Lean artifact can safely support a research claim. |
On native Windows, use the managed Windows runner and the native runtime command target. For Codex-only installs the runtime is usually %USERPROFILE%\.codex\runtime; for multi-agent installs it is usually %LOCALAPPDATA%\ai-agents-skills\runtime. Set $runtime to the installed runtime root, then run:
$runtime = if ($env:AAS_RUNTIME_ROOT) { $env:AAS_RUNTIME_ROOT } elseif (Test-Path "$env:USERPROFILE\.codex\runtime") { "$env:USERPROFILE\.codex\runtime" } else { "$env:LOCALAPPDATA\ai-agents-skills\runtime" }
& "$runtime\run_skill.bat" "skills/lean-strict-verification-gate/run_lean_strict_verification_gate.bat" doctor
PowerShell runner target:
& "$runtime\run_skill.ps1" "skills/lean-strict-verification-gate/run_lean_strict_verification_gate.ps1" doctor
POSIX examples below use run_skill.sh and .sh command targets; use the Windows command target above on native Windows.
Use this skill to prevent overclaiming from generated Lean, skeletons, partial formalizations, or checker output. It separates:
Check the local tool status:
bash ~/.codex/runtime/run_skill.sh \
skills/lean-strict-verification-gate/run_lean_strict_verification_gate.sh doctor
Run non-installing version/toolchain probes when you need reproducibility metadata:
bash ~/.codex/runtime/run_skill.sh \
skills/lean-strict-verification-gate/run_lean_strict_verification_gate.sh doctor --probe
Scan a Lean file without running Lean:
bash ~/.codex/runtime/run_skill.sh \
skills/lean-strict-verification-gate/run_lean_strict_verification_gate.sh scan \
--input formal/final/proof.lean \
--artifact-stage final_candidate
Optionally typecheck only when Lean is already installed:
bash ~/.codex/runtime/run_skill.sh \
skills/lean-strict-verification-gate/run_lean_strict_verification_gate.sh verify \
--input formal/final/proof.lean \
--artifact-stage final_candidate \
--typecheck
For a user-managed Lake workspace, use the explicit Lake environment runner.
The helper requires a project root containing lakefile.lean or
lakefile.toml, records the project context, and still runs the scanner before
typechecking:
bash ~/.codex/runtime/run_skill.sh \
skills/lean-strict-verification-gate/run_lean_strict_verification_gate.sh verify \
--input formal/final/proof.lean \
--artifact-stage final_candidate \
--typecheck \
--runner lake-env-lean \
--project-root /path/to/lean/project
Set AAS_LEAN or AAS_LAKE to select a specific already-installed local
executable. Invalid explicit paths fail closed instead of silently using a
different tool.
The helper never installs Lean, Lake, mathlib, npm packages, Python packages, credentials, services, or MCP servers. Missing Lean reports tool_unavailable.
Before any typecheck, the scanner blocks active:
#evalIO.Processrun_cmdunsafeinitialize@[extern]--allow-importFinal or claim-supporting artifacts also block on active sorry, admit, unsanctioned axiom, unknown trust base, or unreviewed generated proof text. Stubs may contain placeholders only when explicitly marked artifact_stage = stub.
When this skill is involved, consider these workflow templates (install via
the workflow-templates artifact profile, or --with-deps to pull backing skills):
informal-to-lean-formalization-runbook -- Local-first intake mapping an informal proof to Lean declarations with a scanner-first verification gate separating typecheck status from claim support.