The Agent Skills Directory

[PROMPT_INJECTION]: The skill includes instructions that attempt to override the agent's standard behavior and technical perspectives. Specifically, it mandates a specific database session management strategy and requires the agent to refer to the user by a specific name ("The Brougham 22").
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user descriptions to generate persistent files and commit messages, creating an injection surface. 1. Ingestion points: User-provided technical decisions and conversation history (SKILL.md). 2. Boundary markers: Absent; input is interpolated directly into templates. 3. Capability inventory: File system writes to /docs/adr/ and git commit execution (SKILL.md). 4. Sanitization: Absent; no validation or escaping is applied to the input.
[COMMAND_EXECUTION]: The skill requires the use of shell-level git commands to manage documentation history. 1. Evidence: Phase 3 instructions to stage and commit new files (SKILL.md).

create-adr