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.
npx agentmods add skills/7xuanlu/wenlan/verifynpx skills add 7xuanlu/wenlan --skill verifygit clone --depth 1 https://github.com/7xuanlu/wenlanWrote 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.
[](https://agentmods.dev/skills/7xuanlu/wenlan/verify)<a href="https://agentmods.dev/skills/7xuanlu/wenlan/verify"><img src="https://agentmods.dev/badge/skills/7xuanlu/wenlan/verify.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00063 | $0.00804 |
| Opus 5 | $0.00032 | $0.00402 |
| Sonnet 5 | $0.00013 | $0.00161 |
| Haiku 4.5 | $0.00006 | $0.00080 |
Grade A, and why
verify scanned grade A with 1 finding 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 yesterday.
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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
- Daemon HTTP: curl the changed route on the isolated port; recipe in How it starts
The opening of the file, as written. The whole thing — 57 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Verifying wenlan — drive the real surfaces
Launch: use the run-wenlan skill — build, isolated boot, stop, the port
allocation (manual instance :17878, smoke-cli.sh :17881, smoke-mcp.sh
:17882 — they are not interchangeable), and the daemon lifecycle checklist all
live there. Never verify against the shared prod daemon on :7878.
Drive by surface — these scripts ARE the drive recipes (read them for the flow, run them for a full round-trip):
- Daemon HTTP: curl the changed route on the isolated port; recipe in
.claude/skills/prove/references/daemon.md. - CLI:
bash scripts/smoke-cli.sh(capture → memories → search, black-box). - MCP:
bash scripts/smoke-mcp.sh(stdio JSON-RPC initialize → capture → recall).
Gotchas (drive-time):
- Ingest is async (batcher + embedding): poll search up to ~60s before calling a miss a failure.
- Record evidence: prefix any check with
bash scripts/attest.sh. It appends one JSON line to.claude/attest.jsonl— the ledger the weekly sweep audits — and passes the command's exit status through unchanged. If it cannot write the ledger it exits non-zero even when the command passed: an unrecorded run reads to the sweep as "the smoke never ran", so it must never be reported as a pass. (~/.claude/bin/attest.shis a personal macOS helper; it is optional, and it does not exist on Windows or in a fresh checkout. Use the repo script.) WENLAN_NO_AUTOSTART=1on every command that talks to a daemon. Without it a connect failure starts the user's registered background service, and the check then passes against the wrong instance.
Windows (Git Bash)
Both smokes run here as-is; the platform differences are handled inside them via
scripts/lib/host-process.sh. What changes for hand-driven checks:
- No
lsof. Port checks go throughnetstat -ano(seerun-wenlan). A probe that could not run is not a free port — treat a failed probe as fatal, never as "nothing is listening". - No
TMPDIR. Always givemktemp -dan explicit template. - MSYS paths are not daemon paths. Convert with
cygpath -mat the daemon boundary only, into a second variable; the shell keeps the MSYS spelling for its own cleanup. $!is not the OS pid, and killing by pid alone can kill a recycled pid's new owner. Resolve the WINPID and compare the image path before killing.python3may not exist —pythonis the usual spelling.scripts/smoke-mcp.shresolves it; do the same in ad-hoc checks rather than assuming.bash scripts/attest.shworks here, so Windows/verifysatisfies its own evidence contract..claude/skills/prove/SKILL.mdnames the same repo wrapper.
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.
- yesterday Changed · +29 lines 9c60359c845b
- 5d ago First seen · 28 lines · 63 tokens per session scan A 7a9161a9876d
verify is a skill published in the GitHub repository 7xuanlu/wenlan (62 stars, last pushed today), licensed Apache-2.0. It adds 63 tokens to every session and 804 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
shodh-memory
Persistent memory system for AI agents. Use this skill to remember context across conversations, recall relevant information, and build long-term knowledge. Activate when you need to store decisions, learnings, errors, or context that should persist beyond the current session.
braindb-agent
Persistent memory across sessions via the BrainDB agent. Use at conversation start and whenever you need to recall what you know about the user or save new information to long-term memory.
windows-compat
Audit and harden this Rust repo (code-graph-mcp) for Windows correctness: path-spelling drift between producers, the 32,767-char command-line cap, index-key mismatches, and path predicates that assume one ecosystem's layout. Use whenever touching code that builds, compares, prints, or stores a filesystem path; that…
deeprefine
Agent-native DeepRefine refinement loop — same control flow as DeepRefine.refine(), graphify search instead of FAISS, session LLM, dry-run review before approved graph writes.
deeprefine
Agent-native DeepRefine refinement loop — same control flow as DeepRefine.refine(), graphify search instead of FAISS, session LLM, dry-run review before approved graph writes.
remnic-memory-workflow
Shared memory workflow for Claude Code agents connected to Remnic — recall before acting, observe during work, remember at the end. Trigger phrases include "what do you remember about", "save this for later", "any context from last time".