The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as its primary function is to ingest and process untrusted data from arbitrary codebases and document corpora.
Ingestion points: The skill reads file contents, names, and directory structures from user-specified repo_path or corpus_path using Read, Grep, and Glob tools.
Boundary markers: Behavioral boundaries are established in the instructions ("observe and document, never modify"), though no explicit structural delimiters for the data being mapped are defined for the output format.
Capability inventory: The agent has access to Read, Glob, Grep, and Bash tools, providing broad read access to the local filesystem.
Sanitization: No explicit sanitization or filtering of processed file content is mentioned, which is inherent to the goal of structural mapping.
[COMMAND_EXECUTION]: The skill uses the Bash tool to execute standard system utilities for staleness detection and repository mapping.
Evidence: Uses stat -c %Y to check map timestamps and find to identify files modified after a certain date.
Context: These operations are restricted to benign metadata checks and do not involve executing untrusted scripts or piping network data to a shell.

context-mapper