The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process architecture documentation (ARCH-*.md) to drive task creation, which creates a surface for indirect instructions to influence agent behavior.
Ingestion points: The agent reads architecture files from docs/architecture/** as defined in the skill's paths and instructions.
Capability inventory: The agent is authorized to use the Bash(bd:*) tool to execute bd create and bd update commands, and the Write tool to create documents in docs/plans/**.
Boundary markers: No explicit delimiters or instructions are provided to help the agent distinguish between data and potentially malicious instructions embedded within the architecture documents.
Sanitization: There is no specification for sanitizing or validating the content of the architecture documents before it is used to populate shell commands or plan files.

pm-planning