The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from the current conversation and the repository codebase. (1) Ingestion points: The 'conversation en cours' and 'dépôt' (repository) are used as primary inputs (SKILL.md). (2) Boundary markers: Absent. There are no instructions to the agent to delimit or ignore instructions found within the repository or conversation history. (3) Capability inventory: The skill has the capability to write and publish the generated specification to an external project tracking tool. (4) Sanitization: Absent. The synthesized content is published without validation or escaping. This creates a surface where malicious instructions embedded in the code or conversation could influence the agent's output and subsequent automated actions triggered by the 'ready-for-agent' label.

to-spec