The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local bash scripts, specifically discover-projects.sh and engagement-status.sh, to perform discovery of project directories and to retrieve engagement status information. It includes logic to resolve the workspace root and registry paths dynamically through shell commands.
[PROMPT_INJECTION]: The skill processes project-specific data and configuration files to populate a status dashboard, creating a surface for indirect prompt injection where malicious instructions in project files could influence agent behavior.
Ingestion points: consulting-project.json, ~/.claude/cogni-consulting-projects.json, and various files within discovered client project directories.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to ignore or isolate potentially malicious instructions embedded within the processed consulting artifacts.
Capability inventory: The skill is granted extensive capabilities including shell execution via Bash, file system modifications via Write and Edit, and data retrieval through Read, Glob, and Grep tools.
Sanitization: No evidence of sanitization, validation, or escaping of the data ingested from external project files is present before the data is interpolated into the agent's context.

consulting-resume