The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests architectural documentation files (ADRs and ARCH docs) that are processed by the agent and passed as context to a secondary agent (decision-scorer). This creates a vector where malicious instructions embedded in these files could potentially influence the agent's behavior.
Ingestion points: Reads project documentation from docs/adr/, docs/architecture/, and .great_cto/PROJECT.md using the Read and Glob tools.
Boundary markers: The instructions do not define specific delimiters (e.g., XML tags or unique markers) or include specific warnings to ignore embedded commands when parsing the target files.
Capability inventory: The skill utilizes bash for file system querying and the Agent tool for task delegation, which could be leveraged if an injection succeeds.
Sanitization: No filtering or sanitization is performed on the content of the identified documents before it is used as prompt context.
[COMMAND_EXECUTION]: The skill uses the bash tool to execute shell commands (ls, grep, head) for identifying recent documents and counting text patterns. While these are standard file operations, they represent execution of shell logic based on the state of the local file system.

decision-eval