prove

prove is a skill for Claude Code from 7xuanlu/wenlan. It costs 74 tokens per session (553 once invoked), scanned A, original, Apache-2.0.

A verification workflow for Wenlan, with checks for its background service, command-line tool, MCP connection, plugin, test suite, and intended behavior. It records evidence so a later weekly review can assess whether the checks themselves remain reliable.

In plain words
What is it for?
Use it to choose checks based on the files changed, run black-box service or command-line tests, verify MCP and plugin connections, audit tests with mutation testing, and trace tests back to intended behavior.
Why use it?
It catches failures at the boundary between components and checks whether tests can actually detect broken behavior, rather than only confirming that they run.

Skill for Claude Code

Written for Claude Code: installed under .claude/. Also seen: reads .claude/ paths.

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.

agentmods
npx agentmods add skills/7xuanlu/wenlan/prove
Any agent
npx skills add 7xuanlu/wenlan --skill prove
Clone the repo
git clone --depth 1 https://github.com/7xuanlu/wenlan

Made for: Claude Code.

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 prove

README.md
[![agentmods](https://agentmods.dev/badge/skills/7xuanlu/wenlan/prove.svg)](https://agentmods.dev/skills/7xuanlu/wenlan/prove)
Your own site
<a href="https://agentmods.dev/skills/7xuanlu/wenlan/prove"><img src="https://agentmods.dev/badge/skills/7xuanlu/wenlan/prove.svg" alt="Measured on agentmods" height="20"></a>
Per session 74 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 553 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
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 $0.00074 $0.00553
Opus 5 $0.00037 $0.00277
Sonnet 5 $0.00015 $0.00111
Haiku 4.5 $0.00007 $0.00055

Measured 2d ago against content hash af3a2921a8be, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade A, and why

prove 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 2d ago.

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.

.claude/skills/prove/SKILL.md · 34 lines

What it actually says

/prove [surface] — wenlan verification loops

Run every check through the evidence wrapper (records to .claude/attest.jsonl, which the weekly sweep audits): bash scripts/attest.sh <command>. That wrapper ships in the repo and runs on macOS, Linux and Windows (Git Bash); the older ~/.claude/bin/attest.sh is a personal copy that does not exist on every machine, and a missing wrapper means the run is unrecorded, which the sweep reads as "the smoke never ran". Sandboxed local runs need TMPDIR=/tmp/claude prefixed (macOS mktemp gotcha).

Surface Command Proves
daemon read references/daemon.md HTTP behavior on an isolated daemon
cli attest.sh bash scripts/smoke-cli.sh shipped CLI black-box round-trip
mcp attest.sh bash scripts/smoke-mcp.sh stdio JSON-RPC round-trip
plugin /wenlan:setup verify path plugin → MCP → daemon wiring
suite breadth: cargo mutants -f <file> (auto-generated mutants, lib-test oracle); depth: read references/mutprove.md tests can actually go red (mutation audit)
behaviors attest.sh python3 scripts/check-behavior-trace.py <plan.md> <tests...> tests trace to intent, both directions
sweep read references/sweep.md weekly verify-the-verifier

No argument → pick the surfaces the current diff touches (server/core → daemon; cli crate → cli; mcp crate → mcp; new/changed tests → behaviors + suite). App/UI surface (app/, src/, or anything rendered): rendered evidence required — drive the running app and look at the screen; unit/build green is not proof of visible correctness.

Live-smoke legs that need the real GPU model (L7) stay manual: scripts/live-smoke-*.sh — run before merging features whose e2e stubs the LLM.

Files

What ships with it

4 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. 2d ago Changed · +4 lines af3a2921a8be
  2. 6d ago First seen · 30 lines · 74 tokens per session scan A f5ce9c2e9c7f

Subscribe to this mod's changes

prove is a skill published in the GitHub repository 7xuanlu/wenlan (63 stars, last pushed today), licensed Apache-2.0. It adds 74 tokens to every session and 553 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.