Borrowing it
Nothing to install: this file belongs to mtarcure/claude-vibe-squad. 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/mtarcure/claude-vibe-squad/main/.claude/skills/de-ai-before-delivery/SKILL.mdgit clone --depth 1 https://github.com/mtarcure/claude-vibe-squadWrote 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/mtarcure/claude-vibe-squad/de-ai-before-delivery)<a href="https://agentmods.dev/skills/mtarcure/claude-vibe-squad/de-ai-before-delivery"><img src="https://agentmods.dev/badge/skills/mtarcure/claude-vibe-squad/de-ai-before-delivery/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/mtarcure/claude-vibe-squad/de-ai-before-delivery"><img src="https://agentmods.dev/badge/skills/mtarcure/claude-vibe-squad/de-ai-before-delivery.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00086 | $0.00862 |
| Opus 5 | $0.00043 | $0.00431 |
| Sonnet 5 | $0.00017 | $0.00172 |
| Haiku 4.5 | $0.00009 | $0.00086 |
Grade A, and why
de-ai-before-delivery 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 8d 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 — 60 lines — stays where its author put it; the contents beside it link to each section on GitHub.
De-AI Before Delivery
House procedure for the last pass before a deliverable reaches an external reader. Two kinds of tell leak: our internal process, and the machine cadence. Strip both without touching a single technical claim. This runs on the finished artifact at the boundary — not during authoring.
Strip the process tells
- Remove internal phase references (
Phase 3,L5,S2,lane), task IDs, run IDs,_state/and/tmp/paths, worktree paths, and specialist names. A triager who reads "dispatched to exploit-developer in Phase 4" learns about our pipeline, not their bug. - Remove anything that only makes sense inside the squad: mode names, reviewer routing, registry vocabulary, internal file layout, model-lane names.
Strip the AI tells
- Cut the openers and filler: "I'll help you", "Certainly", "Let me", "Great question". Cut hedging stacks ("it may be possible that this could potentially").
- Match the destination's register. Heavy em-dash cadence and symmetrical "not X, but Y" phrasing read as machine-written where the destination is terse; adjust to the register a person who works on that problem actually writes in.
De-AI is a register change, NOT a content reduction
- Keep every technical claim and all of its evidence — reproduction steps, addresses, line numbers, CVSS, PoC. A stripped report that loses the repro is worse than an unstripped one; a triager cannot act on register.
- Do not "clean up" by summarizing away detail. The only things removed are the provenance of how WE work and the cadence of a machine — never what was found or how to reproduce it.
When a campaign ships more than one report, they must not look like a batch
The register pass above works inside one document. A second tell lives across documents, and it is
the one that costs most: several reports from one researcher, landing together, sharing a section
skeleton, each citing file:line and carrying a local harness with no live-service crash. Read at
volume that shape says "automated scan output" before anyone reads the finding, and it shifts the
reader's prior from trust to suspicion on every report in the set — including the good ones.
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.
- 8d ago First seen · 60 lines · 86 tokens per session scan A 5150cae594ae
de-ai-before-delivery is a skill published in the GitHub repository mtarcure/claude-vibe-squad (148 stars, last pushed 2d ago), licensed MIT. It adds 86 tokens to every session and 862 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 2026-09-03.
Other skills, from other repositories
gh-find-prs
Survey open Codewhale PRs and triage each for mergeability and disposition against the real landing branch.
contributor-onboarding
Help a new contributor get productive on this checkout - inspect sync state against main, build, run the repository's exact verification gate, and produce a local what's-new digest. Never fetches, pulls, or modifies a dirty tree on its own. Explicit-only.
codew-release-qa-sweep
Use before claiming Codewhale release work is done: run the full gate sweep and list the manual QA targets.
gh-file-issue
Use when filing a new Codewhale GitHub issue: turn a bug or idea into a well-formed, actionable issue with repro, acceptance criteria, labels, and milestone.
gh-treasure-hunt
Hunt the issue/PR queue for highest value-over-risk wins: clean focused community PRs, already-implemented issues to close, safe quick-fixes.
skill-installer
Install, update, trust, or inspect Codewhale skills from GitHub or local skill folders. Use when the user asks for available skills or wants a community skill installed.