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/forgeyclap/claude-forge/forge-docdriftnpx skills add ForgeyClap/claude-forge --skill forge-docdriftgit clone --depth 1 https://github.com/ForgeyClap/claude-forgeWrote 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/forgeyclap/claude-forge/forge-docdrift)<a href="https://agentmods.dev/skills/forgeyclap/claude-forge/forge-docdrift"><img src="https://agentmods.dev/badge/skills/forgeyclap/claude-forge/forge-docdrift.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 | $0.00051 | $0.01778 |
| Opus 5 | $0.00026 | $0.00889 |
| Sonnet 5 | $0.00010 | $0.00356 |
| Haiku 4.5 | $0.00005 | $0.00178 |
Grade A, and why
forge-docdrift 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 4d 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.
How it starts
The opening of the file, as written. The whole thing — 105 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Forge playbook — Docdrift (external-doc claim drift detection)
Do not duplicate existing tools — this is a NEW, narrow capability. It does not replace forge-scout
(external-capability research/vetting) or forge-audit-loop (this project's own internal health) — it only
watches whether an external page a skill/doc CITES still backs up the claim made about it.
Provenance (honesty)
This tool's IDEA came from researching shanraisshan/claude-code-best-practice (an external community
repo) — the Lead approved adopting the underlying idea ("keep external docs and our own claims about them in
sync") as one of 3 GH-research winners. No code was copied. forge-bin/forge-docdrift.cjs is a
Forge-native reimplementation built from scratch around this project's own conventions (zero-dep,
config/orchestration/*.json seed + .claude/forge-research/* state/ledger, log-event.cjs's existing
audit_finding event type, the same injectable-fetcher hermetic-test seam every other network-touching
forge-bin tool uses). Provenance: PATTERN_ADAPTED (see the global research-recovery policy's honesty
vocabulary) — the pattern was adapted, not the implementation.
What it does
forge-bin/forge-docdrift.cjs check [--source <id>] [--json] [--run <run_id>] reads
config/orchestration/docdrift-sources.json — a seeded list of REAL claims this project's own skills/docs
make that depend on an external, changeable page (an official Claude Code doc page, an n8n node doc page).
For each claim it fetches the real source_url (following redirects, timeout-bounded, zero-dependency
node:https/node:http) and searches the raw response body for every one of that claim's check_tokens.
Every claim is classified against its own persisted history in
.claude/forge-research/docdrift-state.json:
- OK — tokens found, claim was not drifted last known check.
- NEW-DRIFT — a token is missing, claim was NOT drifted before (first time this claim looks stale).
- RECURRING — a token is missing again, same as last known check (a standing, already-reported gap).
- RESOLVED — tokens found again after a prior drift (the doc caught back up, or the claim was fixed).
- UNREACHABLE — the fetch failed/timed out/returned non-2xx. Never counts as drift — the last KNOWN drift state is left exactly as it was. A network hiccup is not proof the docs changed.
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.
- 4d ago First seen · 105 lines · 51 tokens per session scan A f105c0f11876
forge-docdrift is a skill published in the GitHub repository ForgeyClap/claude-forge (2 stars, last pushed 1mo ago), licensed MIT. It adds 51 tokens to every session and 1,778 once invoked, about $0.0003 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 2026-08-31.
Other skills, from other repositories
frame-macos-notification
拟真 macOS 通知 banner + app icon + 标题正文, 适合 video overlay / 产品发布预告.
prototype-web
可点击的功能性 Web 原型, 含导航、英雄区、特性区、CTA.
waitlist-page
【模板: 等候名单页 / Waitlist】 【意图】为新产品 / 早鸟内测做一张极简等候页。 【布局】.
deck-course-module
暖纸背景 + Playfair, 左侧学习目标常驻, 含 MCQ 自测页.
deck-pitch
10 页融资 deck, 白底 + 蓝紫渐变 hero, traction 柱状, $X.XM ask.
argent-tv-interact
Control and inspect TV apps via argent — Apple TV (tvOS), Android TV (leanback), and Amazon Fire TV (Vega). Boot the target, read focus, navigate with the D-pad remote, type, screenshot, and on Vega debug the JS runtime (evaluate, console logs, network inspector). Use when a task targets a TV (runtimeKind "tv", or…