The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill operates by ingesting and processing content from the local codebase, which represents an untrusted data surface. Adversarial content embedded in code comments, markdown files, or documentation could potentially influence agent behavior during exploration.
Ingestion points: The skill utilizes Read, Grep, and Glob tools to import source code and configuration files into the agent's context (e.g., in SKILL.md Phases 1 and 2).
Boundary markers: The instructions do not define explicit delimiters or 'ignore embedded instructions' warnings for the data being analyzed.
Capability inventory: The skill is restricted to read-only search and file access tools (Read, Grep, Glob, ls, find) and does not possess tools for network exfiltration, file system writes, or subprocess execution of untrusted code.
Sanitization: There is no evidence of filtering or escaping logic applied to the ingested codebase content before it is processed by the agent.

code-explore