The Agent Skills Directory

[SAFE]: The skill's primary function is goal clarification and intent shaping, which is implemented through logical instructions and pattern matching. It does not contain any obfuscated code, remote downloads, or unauthorized access patterns.
[PROMPT_INJECTION]: The skill acts as a surface for indirect prompt injection as it ingests untrusted user input and writes it to a persistent database. Ingestion points: User-provided text describing goals or tactics (e.g., in SKILL.md instructions for activating on user phrases). Boundary markers: The instructions do not define specific delimiters or instructions to ignore embedded commands within user input. Capability inventory: The skill utilizes project-sync commands (get-project, update-state) to read and write to a project database. Sanitization: No explicit validation or escaping of user input is specified before data is synced to the database.
[COMMAND_EXECUTION]: The skill executes commands through the project-sync extension. These operations (get-project, update-state, set-stage, add-activity) are used legitimately to maintain the canonical state of a project as described in the skill's operational rules.

intent-shaping