lintai docs

Appearance

Rule Reference

SEC430lintai-ai-securitystablemarkdownwarn

AI markdown: unsafe Edit path grant

AI-native markdown frontmatter grants `Edit(...)` over an unsafe repo-external path

Provider
lintai-ai-security
Surface
markdown
Scope
per_file
Tier
stable
Severity
warn
Confidence
high
Detection
structural
Remediation
message_only

Activation Model

Preset Membership

This rule is part of the builtin activation graph through these preset memberships.

previewmembership
skillsmembership

Lifecycle

Stable Lifecycle Contract

State

stable_gated

Graduation rationale

Checks AI-native frontmatter for `Edit(...)` grants scoped to absolute, home-relative, drive-letter, or parent-traversing paths outside the repository boundary.

Deterministic signal basis

MarkdownSignals exact frontmatter token analysis for `Edit(...)` scopes with absolute, home-relative, drive-letter, or parent-traversing path markers.

Malicious corpus
skill-edit-unsafe-path-allowed-tools
Benign corpus
skill-edit-unsafe-path-allowed-tools-specific-safe
structured evidence required remediation reviewed
Canonical note

Structural stable rule intended as a high-precision check with deterministic evidence.

Nearby Signals

Related Rules

SEC352AI markdown: unscoped Bash tool grantAI-native markdown frontmatter grants unscoped Bash tool accessmarkdownstable
SEC417AI markdown: unpinned pip git installAI-native markdown installs Python packages from an unpinned `git+https://` sourcemarkdownstable
SEC428AI markdown: unsafe Read path grantAI-native markdown frontmatter grants `Read(...)` over an unsafe repo-external pathmarkdownstable
SEC429AI markdown: unsafe Write path grantAI-native markdown frontmatter grants `Write(...)` over an unsafe repo-external pathmarkdownstable
SEC431AI markdown: unsafe Glob path grantAI-native markdown frontmatter grants `Glob(...)` over an unsafe repo-external pathmarkdownstable
SEC432AI markdown: `Bash(git add:*)` tool grantAI-native markdown frontmatter grants `Bash(git add:*)` authoritymarkdownstable

SEC430 / MD-EDIT-UNSAFE-PATH ​

SEC430 flags AI-native markdown frontmatter when allowed-tools or allowed_tools grants Edit(...) over an absolute path, home-relative path, Windows drive path, or parent-traversing path.

Why It Matters ​

Shared instruction frontmatter should not silently authorize edits outside the repository scope. Repo-external or parent-traversing Edit(...) grants weaken reviewability and make workspace boundaries unclear.

Trigger Shape ​

  • AI-native markdown surface with parsed frontmatter
  • path is not fixture-like
  • allowed-tools or allowed_tools contains Edit(...)
  • the inner path starts with /, ~/, ~\\, a Windows drive prefix, or contains ../ or ..\\

Clean Cases ​

  • repo-local scopes such as Edit(./docs/**)
  • frontmatter without Edit(...)
  • fixture-like examples under test or fixture paths

Example Trigger ​

md
---
allowed-tools: Edit(~/workspace/**), Read(./docs/**)
---

Safer Example ​

md
---
allowed-tools: Edit(./docs/**), Read(./docs/**)
---

How To Fix ​

Replace repo-external Edit(...) grants with narrower repo-local scopes, or remove shared edit authority outside the project boundary.