The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-authored data from files like plan.toml and design.md to generate runtime task breakdowns. This process creates a potential surface for indirect prompt injection, as instructions hidden within these project files could influence the agent's planning logic.
Ingestion points: Data is read from plan.toml, design.md, slice-nnn.md, and materialized runtime phase sheets in .doctrine/state/.
Boundary markers: The skill does not define specific delimiters or instructions to treat input file content as passive data rather than executable instructions.
Capability inventory: The skill interacts with the doctrine CLI toolset and triggers a subsequent /execute phase.
Sanitization: No sanitization, escaping, or validation steps for the content of the ingested files are described.
[COMMAND_EXECUTION]: The instructions direct the agent to utilize a set of project management commands (e.g., doctrine slice research, doctrine slice phase). These commands are part of the intended functional scope for managing project states and research within the agent's environment.

phase-plan