Agentpedia

Secret scan report

Reports line numbers where pasted text looks like a token, a cloud key, or a private-key header. Use when the user asks whether a diff, a log, or a note contains a secret.

Look for token-shaped strings in text the user already has. Report the line and the kind. Do not repeat the secret, and do not send the file anywhere.

Steps

  1. Confirm the file or the paste. Refuse a request to scan a machine, a network, or a repository the user did not put in front of you.
  2. Run python3 scripts/scan_text.py on that file. The sample is assets/sample.txt.
  3. For each printed line, record the line number and the kind (aws-access-key, github-token, openai-key, slack-token, private-key, or high-entropy).
  4. Check the count against the findings= line. If the user asks what to do next, say to revoke the credential and to remove it from the file. Do not invent a rotation procedure for a vendor you were not given.

Example

User: Scan assets/sample.txt.

Agent: Line 2 is a github-token. Line 3 is an aws-access-key. The script prints findings=2. The values are not copied into the report. The last line has no finding.

Boundaries

A clean scan is not proof that a secret is absent. The script only matches shapes it knows. Do not claim the file is safe to commit.

References

The shape list follows the detector classes used by gitleaks: known prefixes, private-key headers, and high-entropy tokens. The patterns in scripts/scan_text.py are original and shorter than that project's ruleset.

Lint report

Score 100. 0 errors, 0 warnings, 0 notes.

skill-lint reported no errors, warnings, or notes.

See also

References