Borrowing it
Nothing to install: this file belongs to fold-run/fold. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/fold-run/fold/main/.claude/skills/spec-drift/SKILL.mdgit clone --depth 1 https://github.com/fold-run/foldWrote 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/fold-run/fold/spec-drift)<a href="https://agentmods.dev/skills/fold-run/fold/spec-drift"><img src="https://agentmods.dev/badge/skills/fold-run/fold/spec-drift/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.
<a href="https://agentmods.dev/skills/fold-run/fold/spec-drift"><img src="https://agentmods.dev/badge/skills/fold-run/fold/spec-drift.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.00067 | $0.01189 |
| Opus 5.5 | $0.00027 | $0.00476 |
| Sonnet 5.5 | $0.00013 | $0.00238 |
| Haiku 4.5 | $0.00007 | $0.00119 |
Grade A, and why
spec-drift 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 2mo 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 — 98 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Protocol revision drift
/conformance catches drift in the suite. This catches drift in the
specification — the changes that land months before a check exists for
them, and the deprecations that never get a check at all.
Run it when a revision ships, when the SDK bumps a protocol version, and
before any deliberate CONFORMANCE_COMMIT bump.
Procedure
1. Read the changelog, not the diff.
https://modelcontextprotocol.io/specification/<new>/changelog.md lists
major changes, minor changes, and deprecations. Also fetch
…/deprecated.md — the removal registry, with the twelve-month window
that tells you how long fold has.
2. Classify each entry against fold's two faces. fold is a server to
its clients (gateway/router.go, the mcp.Server and its middleware) and
a client to its upstreams (gateway/upstream.go). A change can hit one,
the other, or both, and the answers differ:
- Server-side changes fold must implement or clients break.
- Client-side changes are usually the SDK's job — check whether
go-sdkalready covers it before writing anything. - Intermediary changes are fold's alone and no SDK will do them: list merging, cache scoping, error-code minting, namespacing. These are the ones that get missed.
3. Grep before concluding. A feature fold "doesn't support" often exists under another name. For each entry, search for the field, method, and header names:
grep -rni 'resultType\|input_required\|cacheScope\|ttlMs\|Mcp-Method' --include='*.go' .
grep -rn '\-320[0-9][0-9]' --include='*.go' . | grep -v _test
4. Check the SDK's position. fold never hand-rolls framing, so for
most wire changes the question is "has the SDK shipped this yet" and the
answer bounds what fold can do. Record the SDK version and the specific
restriction — the subscriptions/listen gap in README "Not implemented"
is exactly this shape, and it carries a drift canary in
gateway/listen_test.go that fails when the SDK lifts its restriction.
A gap that waits on the SDK gets a canary that fails when the wait
ends — otherwise nobody learns the wait is over.
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.
- 2mo ago First seen · 98 lines · 67 tokens per session scan A 2e5471cda1b8
spec-drift is a skill published in the GitHub repository fold-run/fold (1 stars, last pushed 12d ago), licensed Apache-2.0. It adds 67 tokens to every session and 1,189 once invoked, about $0.0003 per session on Opus 5.5. 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.
spec-drift compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| spec-driftfold-run/fold | Skill | 1 | 12d ago | Apache-2.0 | A | 67 | Claude Code |
| robotgoznlgis/opengis-skills | Skill | 76 | 10d ago | MIT | B | 115 | Claude Code, Codex |
| migrate-goldmark-extension-v1-to-v2yuin/goldmark | Skill | 5.1k | 4d ago | MIT | A | 26 | Claude Code |
| migrate-goldmark-app-v1-to-v2yuin/goldmark | Skill | 5.1k | 4d ago | MIT | A | 25 | Claude Code |
| use-modern-goJetBrains/go-modern-guidelines | Skill | 3.9k | 22d ago | Apache-2.0 | A | 32 | Claude Code |
| jwx-guide-v4lestrrat-go/jwx | Skill | 2.4k | 3d ago | MIT | A | 61 | Claude Code |
| goericrisco/rsc-harness | Skill | 503 | 2d ago | MIT | A | 86 | Claude Code, Codex |