The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes Node.js scripts (coupling.mjs, extract-seams.mjs, owners.mjs) that execute the git command-line tool to analyze project history and structure. These scripts use execFileSync with literal argument arrays, which is a secure method that prevents shell injection.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external codebases (such as routes, database schemas, and manifests) to generate node specifications. This creates an attack surface where malicious code in an analyzed project could attempt to influence the agent's reasoning.
Ingestion points: The extract-seams.mjs script performs a filesystem walk and reads project files to build a metadata model.
Boundary markers: The metadata is collected into a structured JSON model (.hsdd-seams.json) before being read by the agent.
Capability inventory: The agent has the capability to write markdown specifications and contract schemas to the hsdd/ project directory.
Sanitization: The scripts do not explicitly sanitize extracted technical strings (e.g., route paths, table names) for natural language instructions before they are processed by the agent.
[SAFE]: The extract-seams.mjs script implements path validation when writing its JSON model, ensuring that output files cannot be written outside the intended project root or current working directory.

hsdd-adopt