The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes a local shell script (templates/resolve-config.sh) and standard command-line tools (git, jq, grep) to resolve project-specific configurations and retrieve historical information from the git log. These operations are performed locally and are essential to the skill's core functionality.
[DATA_EXFILTRATION]: The skill includes a dedicated security guardrail that explicitly instructs the agent to scan for and exclude secrets from any generated documentation. No unauthorized network operations, data exfiltration patterns, or connections to external domains were identified.
[PROMPT_INJECTION]: The skill manages the risk of indirect prompt injection inherent in documentation tasks through specific instructional guardrails.
Ingestion points: The skill reads existing Markdown documentation, repository metadata (e.g., package.json), and external architectural decision records (ADRs) during the reconciliation process.
Boundary markers: The system relies on YAML frontmatter delimiters and standardized Markdown headers to separate data from instructions.
Capability inventory: The skill has access to file system operations (Read, Write, Edit) and local command execution (Bash).
Sanitization: Mandatory verification of all documentation facts against the project's source code, secret detection scans, and restricted "re-homing" of existing prose (moved, never rewritten) serve as primary mitigations.

write-docs