.gitleaks.toml

# gitleaks configuration for cli-agent-orchestrator (issue #457).
#
# Extends gitleaks' built-in rule set via `useDefault` so we stay current with
# upstream credential detections rather than maintaining our own copy, adds two
# custom rules for shapes still missing from the defaults, and allowlists one
# well-known documentation placeholder.
#
# This complements — it does not replace — the runtime redactor in
# secret_gate.py (which guards memory writes / archive export at runtime). This
# config gates git content in CI so a real credential can't land on a branch.
#
# False positives: the scan keys on real credential SHAPES (AWS/GitHub/GitLab
# keys, PEM private keys, JWTs, Slack webhooks), not on placeholders, env
# references, hashes, UUIDs, or the word "secret". Ordinary code/docs/fixtures
# pass. The one realistic edge case for this repo is a provider fixture that
# captures a JWT-shaped token from a live CLI banner. If that happens and the
# token is NOT a real credential, either scrub it (preferred — see the fixture
# hygiene guidance) or exempt that single line with a trailing `# gitleaks:allow`
# annotation. Do NOT broadly allowlist `test/providers/fixtures/**` — that would
# blind the gate to a real key captured into a fixture, which is exactly the
# incident (#436) this guards against.
title = "CAO gitleaks config"

[extend]
useDefault = true

# Custom rules closing gaps still present in gitleaks' default set (verified
# missing as of gitleaks 8.30.1). These credential shapes are real and worth
# catching.
[[rules]]
id = "aws-secret-access-key"
description = "AWS secret access key (40-char base64) in an assignment"
# Anchored to an aws-ish key name so we match the secret in context rather than
# flagging every 40-char base64 blob (which would false-positive on hashes).
# The trailing (?:[^A-Za-z0-9/+]|$) requires the 40-char run to END at a
# non-base64 char OR end-of-line, so a key terminated by ; , ) & etc. (common in
# code/shell) is still caught, while a 41st base64 char correctly disqualifies
# it (real AWS secrets are exactly 40).
regex = '''(?i)aws.{0,20}(?:secret|access).{0,20}['"=:\s]([A-Za-z0-9/+]{40})(?:[^A-Za-z0-9/+]|$)'''
keywords = ["aws"]

[[rules]]
id = "github-fine-grained-pat"
description = "GitHub fine-grained personal access token (github_pat_...)"
# Structure-accurate to avoid flagging doc placeholders like
# `github_pat_..._example_placeholder`: the real format is `github_pat_` + 22
# alphanumerics + `_` + 59+ alphanumerics. Requiring that shape (not a loose
# [0-9a-zA-Z_]{22,255}) rejects underscore-laden placeholders while catching
# real tokens.
#
# This id matches a built-in rule, so `useDefault` merges the two; we set
# `entropy` explicitly (rather than inheriting the built-in's 3.0) so a
# low-entropy placeholder that happens to fit the shape is still rejected and
# the threshold is visible here instead of implied.
regex = '''github_pat_[0-9a-zA-Z]{22}_[0-9a-zA-Z]{59,}'''
entropy = 3.0
keywords = ["github_pat_"]

# Array-table form ([[allowlists]], plural) rather than the singular [allowlist],
# which gitleaks has deprecated for removal in v9.
[[allowlists]]
description = "Intentional non-secrets."

# We deliberately do NOT allowlist whole test files, PEM key material, or whole
# commits — each of those would suppress real findings too. The synthetic PEM
# sample in test/services/test_secret_gate.py has only 16 chars of key body,
# below gitleaks' private-key rule threshold (>=64), so it isn't flagged and
# needs no allowlist. The only allowlisted value is AWS's documented example
# access key: the exact 20-char string, word-bounded so a real key merely
# containing it as a substring is NOT exempt.
#
# For any genuine historical false positive, prefer a finding-level
# .gitleaksignore fingerprint over a commit- or path-level allowlist.
regexes = [
  '''\bAKIAIOSFODNN7EXAMPLE\b''',
]