The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes git commands via subprocess.run to establish repository identity and manage the local ignore boundary (.git/info/exclude). These operations are performed without a shell and are confined to the repository root.
[COMMAND_EXECUTION]: The skill protocol runtime facilitates the execution of the skill's internal validation script, seal_blueprint.py. Arguments passed to this script are defined in the skill manifest and undergo strict path resolution and validation to prevent directory traversal or unauthorized access.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process codebase information and markdown-based plan drafts, creating a surface for potential indirect instructions.
Ingestion points: The blueprint_runtime.py script reads the contents of request.md and blueprint.md from the local filesystem.
Boundary markers: SKILL.md contains explicit instructions for the agent to treat repository content as untrusted evidence rather than authority.
Capability inventory: The skill possesses the capability to write task artifacts to the blueprints/ directory, modify Git ignore rules, and execute repository-scoped commands.
Sanitization: The runtime implements rigorous path sanitization, including checks for repository containment, symlink resolution, and the blocking of relative path traversal. It also enforces document integrity using SHA256 hashing and structural validation.

blueprint