agent-desktop/.gitleaks.toml

title = "agent-desktop secret and privacy scanning"

# Inherit gitleaks' built-in credential rules (API keys, tokens, private keys),
# then add project-specific privacy rules below. This repo is public and ships
# planning documents, probe evidence, and traces that are written on developer
# machines, so operator identity and filesystem layout are treated as leaks
# alongside conventional secrets.
#
# Regexes are Go RE2: no lookahead, no lookbehind, no backreferences. Anything
# that would need a negative lookahead is expressed as an allowlist entry.
[extend]
useDefault = true

[[rules]]
id = "ad-home-directory-path"
description = "Absolute home-directory path exposing an operator account name"
# The first character after the home root must not be a dot, so relative
# fragments like "$suite_root/home/.cache" are not mistaken for "/home/<user>".
regex = '''(?i)(/Users/|/home/|[A-Za-z]:\\Users\\)[A-Za-z0-9_-][A-Za-z0-9._-]{1,}'''
tags = ["privacy", "path", "identity"]

[[rules]]
id = "ad-windows-account-sid"
description = "Windows account SID with machine-unique sub-authorities"
regex = '''S-1-5-21-[0-9]{5,}-[0-9]{5,}-[0-9]{5,}-[0-9]{3,}'''
tags = ["privacy", "identity"]

[[rules]]
id = "ad-personal-email"
description = "Personal email address in tracked content"
regex = '''[A-Za-z0-9._%+-]+@[A-Za-z0-9.-]+\.[A-Za-z]{2,}'''
tags = ["privacy", "identity"]

[[rules]]
id = "ad-machine-hostname"
description = "Developer machine hostname captured from a probe or trace"
# Case-sensitive on purpose: Windows default hostnames are uppercase
# (DESKTOP-A1B2C3D). A case-insensitive form matches this project's own crate
# names (agent-desktop-core, agent-desktop-ffi, ...) and is pure noise.
regex = '''\b(DESKTOP|LAPTOP)-[A-Z0-9]{7}\b'''
tags = ["privacy", "identity"]

# ---------------------------------------------------------------------------
# Allowlist: documentation placeholders, generated artifacts, and OS vocabulary.
# The rules above intentionally match broadly; these entries keep the scan
# usable on content that only *looks* like a leak.
# ---------------------------------------------------------------------------
[allowlist]
description = "Documentation placeholders and generated artifacts"
# Match allowlist regexes against the whole line, not just the captured secret,
# so a placeholder is recognised from its surrounding context.
regexTarget = "line"
paths = [
  '''Cargo\.lock''',
  '''package-lock\.json''',
  '''(^|/)tests/fixtures/''',
  # Never-committed local state and build output. Listed so a local
  # `gitleaks dir .` scan matches what CI sees on tracked content.
  '''(^|/)target/''',
  '''(^|/)node_modules/''',
  '''(^|/)\.claude/''',
]
# Probe captures under probes/*/captures/ are deliberately NOT allowlisted:
# they are UIA tree dumps taken on a developer desktop and are the most likely
# place for a redaction failure, so gitleaks is the backstop behind the probe
# corpus's own redaction gate.
regexes = [
  # Explicit redaction markers and angle-bracket placeholders.
  '''<redacted[^>]*>''',
  '''S-1-5-21-<[^>]*>''',
  # Generic doc examples that teach path shape rather than expose one.
  '''(?i)/Users/(name|user|username|you|youruser|me|someone)''',
  '''(?i)/home/(name|user|username|you|youruser|runner|me|someone)''',
  '''(?i)[A-Za-z]:\\Users\\(name|user|username|you|youruser|runner)''',
  # CI runner home paths are not operator identity.
  '''/Users/runner''',
  '''/home/runner''',
  # Well-known localized group principals are OS vocabulary, not identity.
  '''(?i)BUILTIN\\Admin[a-z]*''',
  # Addresses that are meant to be public.
  '''(?i)[A-Za-z0-9._%+-]+@example\.(com|org|net)''',
  '''(?i)noreply@''',
  '''(?i)users\.noreply\.github\.com''',
]