The Agent Skills Directory

[SAFE]: The skill performs localized file operations to manage legal playbook entries in Markdown format. No evidence of malicious behavior, exfiltration, or unauthorized access was detected.
[COMMAND_EXECUTION]: The skill utilizes local Python scripts (detect_conflicts.py, discover_playbook.py, validate_schema.py) to maintain playbook integrity. These scripts perform benign tasks such as checking for duplicate clause IDs, validating Markdown frontmatter against a JSON schema, and locating the playbook directory within the project structure. These operations are conducted using the 'uv run' command and standard libraries.
[EXTERNAL_DOWNLOADS]: The bootstrap-mode protocol references a seeding process that extracts a bundled baseline tarball containing initial legal clauses. This process includes a verification step where the SHA256 hash of the content is checked against a manifest, ensuring the integrity of the bundled assets.
[DATA_EXPOSURE]: File system access is restricted to the legal-playbook/ directory and project-related files. The skill does not attempt to access sensitive directories such as SSH keys, environment variables, or cloud provider credentials.
[INDIRECT_PROMPT_INJECTION]: The skill ingests user input, including contract snippets, to assist in authoring clauses. While this represents an ingestion point for external data, the skill's functionality is limited to generating static Markdown files and performing structural validation, which prevents the execution of malicious instructions embedded in the analyzed text.

legal-playbook-author