The Agent Skills Directory

[PROMPT_INJECTION]: The "Post-Completion: Workflow Continuation" section contains instructions to bypass user interaction. Specifically, it states "Do NOT prompt the user for confirmation: this is a lightweight checkpoint, not an interactive gate." This instruction attempts to override the agent's default safety behavior of confirming actions with the user.
[INDIRECT_PROMPT_INJECTION]: The skill processes project specifications to generate planning documents and writes to the file system, which creates a vulnerability to instructions hidden in the processed data.
Ingestion points: The skill ingests user-provided specifications at runtime to convert them into implementation plans.
Boundary markers: None. The skill does not define delimiters or specific instructions to disregard embedded commands within the input text.
Capability inventory: The skill is capable of writing to the local file system (e.g., docs/implementation-plan.md, docs/tradeoffs.md) and invoking additional skills such as Skill(attune:project-execution).
Sanitization: None. There are no defined mechanisms for filtering, validating, or escaping content found within the input specification documents.

project-planning