The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill fetches and reads content from external sources, including full bodies and comments from URLs or issue numbers provided by the user (Step 1). This data is then used to generate output (tickets/issues). The instructions lack explicit boundary markers or sanitization steps to prevent instructions embedded in those external sources from influencing the agent's behavior during the ticket-drafting process.
Ingestion points: SKILL.md (Step 1: Gather context
fetches full body + comments from user-provided URLs/references).
Boundary markers: Absent. The skill does not instruct the agent to ignore or delimit potentially malicious instructions within the fetched context.
Capability inventory: The skill has the ability to write new issues to the project tracker (Step 5) and interact with the codebase via glossary tools.
Sanitization: Absent.
[DYNAMIC_EXECUTION]: The skill utilizes uvx to execute the disambiguate tool at runtime (Step 2: Explore codebase). uvx (from the Astral uv project) dynamically downloads and runs Python packages, which constitutes dynamic execution of external code.
[EXTERNAL_DOWNLOADS]: The skill references an upstream skill from a GitHub repository (github.com/mattpocock/skills) for its derivation. This reference is documented neutrally as a source for the skill's logic.

to-tickets