The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection via project-specific documentation and technical specification files.
Ingestion points: The agent reads content from various local files such as .monkeycode/specs/{FEATURE_NAME}/design.md, tasklist.md, and requirements.md to determine its next steps.
Boundary markers: The skill lacks explicit boundary markers or instructions that would cause the model to ignore potential commands embedded within the design or requirement documents.
Capability inventory: The skill has significant capabilities including reading and writing to the file system, executing shell commands (to run tests), and invoking other platform skills (e.g., project-wiki).
Sanitization: There is no evidence of sanitization or filtering applied to the content of the documents before the agent processes them.
[COMMAND_EXECUTION]: The skill uses shell commands to interact with the project environment and verify implementations.
Evidence: The workflow explicitly uses cat to read documentation and instructs the agent to "Run tests" and "Execute all tests," which involves running arbitrary code or test runners in the shell environment.

feature-implementer