The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill relies on reading and processing various project-specific files which could potentially contain malicious instructions if the project environment is compromised.
Ingestion points: The skill reads configuration and design data from .styleseed/project.json, .styleseed/artifacts/index.json, .styleseed/bundles/<artifact-id>.md, .styleseed/manifests/<artifact-id>.json, and STYLESEED.md.
Boundary markers: No specific delimiters or instructions to ignore embedded commands within these files are provided.
Capability inventory: The skill has access to Bash, Write, Edit, and WebFetch tools, and it executes shell commands.
Sanitization: There is no mention of sanitizing or validating the content of the project files before they are used to guide code generation or script execution, although the skill mentions using source hashes to detect drift.
[DYNAMIC_EXECUTION]: The skill executes a local script to resolve context, which is a common pattern for development tools but involves runtime execution of code.
Evidence: Step 1 instructions include running scripts/resolve-context.mjs --project-root . --from-lock STYLESEED.md --agent <agent> via the Bash tool.
[COMMAND_EXECUTION]: The skill uses the Bash tool to execute project-specific scripts and potentially other slash commands (/ss-setup, /ss-resolve, /ss-score, /ss-verify) provided by the agent environment.

ss-build