- Home
- Skills
- Code Quality & Review
- Public Repo Hygiene Check
Public Repo Hygiene Check
Review tracked content, sensitive filenames, assets and reachable Git metadata before public release. Get redacted candidates, contextual false-positive exclusions and a clear remediation/coverage checklist.
$8
Works with the AI Agents you already use
Public Repo Hygiene Check
Example session with this skill installed
Use public-repo-hygiene-check on the public harness snapshot and the explicitly synthetic fixture. Produce separate redacted reports. Review HEAD tracked text, filenames and reachable Git metadata. Do not open environment files, execute repository code or change either repo. Historical blob scanning and binary metadata inspection are excluded and must be disclosed.
Codex followed SKILL.md in this session using read-only commands and tools/capture_repo_evidence.py. The helper is local construction tooling, not included in the buyer ZIP and not a comprehensive secret scanner. JSON files preserve the collected redacted evidence. The synthetic repo has one locally created commit and no remote. All eight seeded categories are listed in fixture-report.md.
- Read your context and instructions
- Compiled the public repo hygiene
- Generated the document
These are agent-authored outputs from a real review of the labeled inputs, not a bundled script's automatic result, independent validation or a promised future outcome.
Public-repository hygiene review — synthetic fixture
Scope and result
Run on the synthetic fixture created for this demo, revision 2fff4d9ce8fe77d28cef2dcc6c3a2cefb2494143, 2026-11-14. Four tracked files, clean working tree, one reachable commit. No untracked/ignored files reported. Read allowed text, inventoried excluded filenames, examined Git metadata. All eight seeded review categories were identified. This demonstrates review behavior, not real credential exposure or publication safety.
Findings and seeded-category coverage
| ID | Seeded category | Evidence | Classification / priority | Suggested action |
|---|---|---|---|---|
| R1 | Local machine path | config.txt:1, value [REDACTED] | Deliberate synthetic example / Low here | Use a relative path or documented placeholder |
| R2 | Contact email | config.txt:2, value [REDACTED] | Synthetic contact / Low here | Remove personal contact unless approved for publication |
| R3 | Token-shaped string | config.txt:3, value [REDACTED] | Fake credential indicator / High if real | Owner would rotate a real exposed key before cleanup; never test it |
| R4 | Private-client reference | config.txt:4, value [REDACTED] | Explicit fictional label / review candidate | Replace confidential identifiers with neutral examples |
| R5 | Tracked environment file | .env, filename only; content not read | Exposure candidate / High if real secrets | Owner reviews privately; untracking and ignore rules do not erase history |
| R6 | Portrait asset candidate | portrait.svg, filename only | Requires asset review / Medium | Confirm image content, rights and metadata; filename alone does not establish a real face |
| R7 | Author/committer contact in history | commit 2fff4d9, identities [REDACTED], non-noreply | Synthetic identity / review candidate | Agree approved public identity before any separate history work |
| R8 | Session marker/link in commit message | commit 2fff4d9, value [REDACTED] | Deliberate synthetic session link / review candidate | Remove sensitive message content only under an approved history plan |
Limits
No environment contents, historical blobs, asset visual/metadata inspection, credentials, remotes, release artifacts or caches were inspected. Asset R6 is correctly a review candidate, not a claim that a person's photograph was detected. Regex results cannot determine confidentiality without context. Nothing was rewritten, deleted or pushed.
Remediation checklist
- Owner distinguishes genuine secrets from explicit examples.
- Rotate real exposed keys, then remove public values and inspect reachable history.
- Replace local paths and confidential references; approve contact identity.
- Review environment filenames and image metadata before publication.
- Review message history and public caches under a separate approved plan.
Evidence: fixture-evidence.json; fixture source and local Git history remain beside this report. No guarantee that the repository is safe to publish.
Public-repository hygiene review — public Harness
Scope and result
Run on public harness at 96f9fcee9453c6bec98026f40f76ed426bad3e6e, 2026-11-14. HEAD inventory: 162 tracked files, clean working tree; no untracked or ignored names reported. Reviewed supported tracked text and metadata of 211 reachable commits. Evidence collector output is in harness-evidence.json. This is not a full historical-content scan.
Findings and classification
| ID | Evidence | Classification | Priority / action |
|---|---|---|---|
| H1 | AGENTS.md:18 and CLAUDE.md:10 match a session marker | Intentional rule prohibiting such markers, not a leaked session | No remediation; retain contextual exclusion |
| H2 | 158 commit messages contain coauthor email metadata; values withheld | Public tool coauthor attribution, domain anthropic.com; publication-policy review candidate | Low; decide whether the repo's no-email rule excludes these public attributions. No evidence here of private customer identity |
| H3 | .env.example tracked; contents excluded | Expected example filename, not a demonstrated credential | Owner may review privately; no claim its content is safe |
| H4 | docs/images/agents.png, galaxy.png, home.png, plan.png | Assets requiring visual and metadata review | Review screenshots before release; no claim they contain people or secrets |
All 211 author/committer identities use the GitHub noreply domain. No session link/marker matches were detected in the inspected commit messages. The two HEAD pattern hits are rules, not actionable leaks. No token-shaped or local-path match was detected by the collector in inspected text, but pattern coverage is narrow and does not guarantee absence.
Not evaluated
Historical blobs and deleted secrets, image contents/metadata, LICENSE (unsupported extensionless text in this collector), environment example contents, LFS objects, unadvertised refs, remote issues/releases/caches, encoded secrets and business identifiers that lack explicit confidential labels. Tool/provider brands in source are not automatically confidential client names. The local clone contains only publicly fetched material; it does not represent the author's private working tree.
Remediation checklist
- Confirm whether public coauthor attribution is permitted under the repo policy.
- Review screenshots and example configuration in an appropriate scoped pass.
- Add historical-content and publication-surface review before any assurance of a clean release.
- Preserve redacted evidence; do not rewrite history or push without a separate decision.
Result: no confirmed private-data disclosure in the inspected surfaces; one policy-review candidate and the explicit coverage gaps above. No guarantee of publication safety. Source conventions: public AGENTS.md and CLAUDE.md; this review made no repository changes.
agents.png
PDF · document
Example file from a real run - the skill writes it into your workspace.
Connects securely to your tools. The creator never sees your data.
About this skill
A Markdown review workflow for personal-data and secret indicators, local paths, private references, session links, assets and Git identities. It distinguishes intentional examples from possible disclosures and records exactly which surfaces were inspected. Includes SKILL.md only. The demos used local construction tooling that is not included in the buyer package. No automatic scanner, secret validation, cleanup or history rewrite is supplied.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Code Quality & Review.
Requires read-only file access and local git. Cannot validate if credentials are active. Does not inspect LFS objects or remote caches. Does not rewrite git history.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 2 days ago