The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of the canon and orchestra CLI tools to query and modify a local SQLite database. It executes commands such as canon list, canon create, and canon supersede to manage the lifecycle of ideas and areas.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted record content (subjects, bodies, and sources) from the canon store and displays it to the agent, which could lead to malicious instruction execution if a record contains adversarial content. * Ingestion points: SKILL.md Step 2 and 4 (listing and displaying ideas). * Boundary markers: Absent; the instructions do not specify using delimiters or warnings when rendering external idea content. * Capability inventory: SKILL.md contains multiple database write operations (create, supersede) and state management tools (orchestra). * Sanitization: Absent; there is no mention of filtering or escaping the contents of the ideas before processing.
[DYNAMIC_EXECUTION]: In Step 6 of SKILL.md, the skill generates an 'area body' as a local file (<area-body.md>) and then consumes it as a file-based argument for the canon create command, which is a form of local script/content generation and execution.

triage-ideas