The Agent Skills Directory

[COMMAND_EXECUTION]: The skill describes a 'Small demo / spike' workflow where the agent is encouraged to write and execute code (e.g., spike.py, try-{topic}.ts) for 5-30 minutes to verify the behavior of APIs or libraries.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by reading and summarizing data from project files such as AGENTS.md and various files within the codestable/ directory.
Ingestion points: Reads AGENTS.md and performs glob/grep operations on codestable/ documentation, features, and roadmaps.
Boundary markers: Absent; there are no instructions to the agent to treat external file content as untrusted data or to use delimiters.
Capability inventory: The agent can write markdown files to the repository and execute shell commands to run spike code.
Sanitization: No sanitization or verification of file content is performed before the agent processes and acts upon the information.
[COMMAND_EXECUTION]: The skill performs automated file system management, including directory discovery via globbing and creating new documentation files in specific project paths like codestable/features/ and codestable/brainstorms/.

cs-brainstorm