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 skills add joshuafuller/claude-spill-scrub --skill spill-scrubgit clone --depth 1 https://github.com/joshuafuller/claude-spill-scrubWrote 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/joshuafuller/claude-spill-scrub/spill-scrub)<a href="https://agentmods.dev/skills/joshuafuller/claude-spill-scrub/spill-scrub"><img src="https://agentmods.dev/badge/skills/joshuafuller/claude-spill-scrub/spill-scrub/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/joshuafuller/claude-spill-scrub/spill-scrub"><img src="https://agentmods.dev/badge/skills/joshuafuller/claude-spill-scrub/spill-scrub.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 | $0.00102 | $0.01507 |
| Opus 5 | $0.00051 | $0.00754 |
| Sonnet 5 | $0.00020 | $0.00301 |
| Haiku 4.5 | $0.00010 | $0.00151 |
Grade A, and why
spill-scrub 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 12d 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 — 132 lines — stays where its author put it; the contents beside it link to each section on GitHub.
spill-scrub
Claude Code will tell you that you just pasted a live token. It does not clean up after itself. The token stays in the session transcript on disk and is read back into context every time that session is resumed. This skill does the cleanup.
The rule that governs everything here
Scrubbing is not remediation. Deleting a key from a log does not revoke the key. Every run ends by telling the user to rotate. Never let a user finish this workflow believing the incident is closed because the files are clean.
Running the script
spillscrub.py sits next to this file. Always invoke it as:
python3 ${CLAUDE_SKILL_DIR}/spillscrub.py <args>
${CLAUDE_SKILL_DIR} resolves to this skill's own directory whichever way it was
installed — plugin, ~/.claude/skills/, or a project's .claude/skills/ — so do
not guess at paths or cd first.
This only makes sense on the machine whose logs you are cleaning. In a cloud or
Cowork session it would scan that container's ~/.claude, not the user's. If the
session is not local, say so and stop.
scan is pre-approved in this skill's frontmatter and runs without a permission
prompt. scrub is deliberately not pre-approved: it rewrites files in place,
so it should cost a permission prompt every time.
Everything below writes spillscrub.py for brevity.
Workflow
1. Scan first. Always.
python3 ${CLAUDE_SKILL_DIR}/spillscrub.py scan --context --manifest ~/spill-manifest.json
Scan is read-only. It takes about 15 seconds over a 600 MB corpus on a many-core
box. Never jump straight to scrub.
2. Read the two tiers differently
- CERTAIN (tier 1) — vendor-prefixed credentials (
sk-ant-,ghp_,AKIA,glpat-, private key blocks, passwords inside URLs). Treat every one as a real leaked credential. Do not ask the user to confirm these one by one. - REVIEW (tier 2) — shape-based hits (
DB_PASSWORD=…,sshpass -p …,Authorization: Bearer …). These carry false positives. Use--contextto show the masked surrounding text and triage them with the user. The context string never contains the value.
What ships with it
2 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.
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.
- 12d ago First seen · 132 lines · 102 tokens per session scan A e7f684dd4906
spill-scrub is a skill published in the GitHub repository joshuafuller/claude-spill-scrub (1 stars, last pushed 16d ago), licensed MIT. It adds 102 tokens to every session and 1,507 once invoked, about $0.0005 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
binary-analysis
Analyze binary files (exe, dll, sys, bin, ocx, scr, cpl, drv, elf, so, macho, apk) to assess if they are malicious, perform decompilation, extract strings/imports/exports, detect malware, and provide threat assessment. Use this skill when user asks to analyze, examine, check, or assess any binary file, asks if a file…
secrets-gitleaks
Hardcoded secret detection and prevention in git repositories and codebases using Gitleaks. Identifies passwords, API keys, tokens, and credentials through regex-based pattern matching and entropy analysis. Use when: (1) Scanning repositories for exposed secrets and credentials, (2) Implementing pre-commit hooks to…
kernel-cve-analysis
Query the Android/AOSP kernel CVE database to look up a specific CVE, find CVEs affecting a kernel version or build date, find unpatched CVEs in a branch, or identify exploitable vulnerabilities. Use this skill when the user asks about Android kernel CVEs, AOSP kernel vulnerabilities, which CVEs affect a kernel…
provisioning-wizard
Bash wizard for human-only setup.
ai-change
Use this skill whenever the user wants change-management paperwork (RFC, change request, change record, governance doc, rollback doc, CAB submission) for anything AI or ML. This is the correct skill — not a generic RFC skill — any time the subject of the change is a model, LLM, chatbot, AI agent, prompt, RAG pipeline…
change-request
ITIL change request and CAB review workflow. This skill should be used when the user asks to "create a change request", "write an RFC", "plan a change", "CAB review", "change management", "schedule maintenance", "server migration plan", "upgrade plan", "AI model deployment", "GitOps change", or describes any planned…