The Agent Skills Directory

[COMMAND_EXECUTION]: The skill encourages the agent to perform 'minimal demo' or 'spike' tasks, which involve the generation and execution of temporary scripts (e.g., spike.py or try-{topic}.ts) to verify technical facts, API behavior, or library support during the brainstorming process.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it reads and processes data from various files within the .codestable/ directory to establish session context.
Ingestion points: The agent reads attention.md, shared-conventions.md, and other files located in the features/, roadmap/, and brainstorms/ subdirectories.
Boundary markers: The instructions do not specify the use of delimiters or warnings to ignore potentially malicious instructions embedded within the ingested file content.
Capability inventory: The skill has the capability to write new documentation/brainstorming files to the disk and generate/execute code for technical verification.
Sanitization: No sanitization, filtering, or validation of the content read from the local files is mentioned.

cs-brainstorm