The Agent Skills Directory

[DATA_EXPOSURE]: The skill accesses .env and .env.example files during its validation phase. This access is specifically scoped to cross-referencing port definitions against those found in documentation. Since the skill has no network tools allowed and lacks exfiltration patterns, this access is considered safe for its intended purpose.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from markdown files throughout the codebase to identify documentation drift. The risk of command injection is mitigated by the use of highly restrictive regular expressions (e.g., [a-zA-Z0-9_./-]+\.(ts|js|py|rb|go|rs|java|tsx|jsx)) that ensure only valid code identifiers and paths are processed.
[COMMAND_EXECUTION]: The skill utilizes Bash to run discovery and validation commands including find, grep, and git log. All commands are hardcoded in the skill and operate strictly on local file metadata and content to report documentation status.
[METADATA_POISONING]: The skill's configuration and frontmatter are consistent with its described behavior. The allowed-tools list correctly restricts the environment to non-networked utilities, supporting the 'read-only' claim.

scan