The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute local scripts such as scripts/gate_pathspec_check.sh and scripts/ablation_calibrate.sh, and shell utilities like grep and diff for checking path coverage and resolving pointers between assets.
[INDIRECT_PROMPT_INJECTION]: The skill processes content from other context assets like SKILL.md and CLAUDE.md. 1. Ingestion points: Reads target context assets during the diagnosis and splitting phases as described in SKILL.md Step 1. 2. Boundary markers: No explicit delimiters or instructions to ignore embedded content are specified for the processed files. 3. Capability inventory: The agent is granted Bash, Edit, and Write capabilities which are used in conjunction with file processing. 4. Sanitization: No validation or filtering of instructions within the ingested text is performed before processing.

salience-splitter