The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface through its reliance on external project data.
Ingestion points: The skill reads various JSON files from the project workspace, including portfolio.json, products/*.json, features/*.json, markets/*.json, customers/*.json, and propositions/*.json to derive consulting insights.
Boundary markers: No specific delimiters or instructions to ignore embedded commands within these data files are provided in the skill instructions.
Capability inventory: The skill can modify project files using Write/Edit tools, execute local scripts via Bash, and initiate specialized research tasks using the Agent tool.
Sanitization: There are no explicit validation or sanitization steps mentioned for the data ingested from the project JSON files.
[COMMAND_EXECUTION]: The skill executes local bash scripts to perform structural project operations.
Utility Script Invocations: It uses scripts such as cascade-rename.sh, validate-entities.sh, and project-status.sh to maintain consistency across the project portfolio.
Dynamic Path Resolution: The skill resolves the location of these scripts using a shell expansion pattern (${CLAUDE_PLUGIN_ROOT:-$(ls -td "$HOME"/.claude/plugins/cache/insight-wave/cogni-portfolio/*/ | head -1)}) to ensure compatibility with various installation environments.

markets