The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes shell commands including find, grep, cat, ls, and jq to inspect the local filesystem and extract metadata from project files such as package.json and tsconfig.json. This behavior is expected given the skill's purpose of codebase auditing.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it reads and processes untrusted data from the user's codebase.
Ingestion points: Extensive reading of source code files across the repository in Step 5 and Step 6.
Boundary markers: Absent; the instructions do not use delimiters or provide warnings to the agent to distinguish codebase content from its own instructions.
Capability inventory: The agent can execute shell commands and write to files (specifically creating .candid/Technical.md and .candid/config.json).
Sanitization: None; contents from the analyzed code are processed directly by the agent to synthesize standards.

candid-init