The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from project files and external sources, creating a surface for indirect prompt injection.
Ingestion points: Reads MARKETING.md from the project root and potentially fetches content from user-provided domains via the agent_readiness_lint.py script.
Boundary markers: The skill lacks explicit delimiters or instructions to ignore embedded commands within the ingested data.
Capability inventory: The skill has broad file-writing permissions within the workspace/ directory and executes several local Python scripts. It also directs the agent to invoke other high-capability skills (e.g., social posting, email sending) based on the generated output.
Sanitization: No validation or escaping of external content is specified before it is incorporated into the planning process.
[COMMAND_EXECUTION]: The skill performs shell execution of local Python scripts to validate generated content.
Evidence: Execution of python scripts/quality_gates/banned_word_check.py, four_us_score.py, seo_lint.py, and agent_readiness_lint.py.
Context: These scripts are part of the internal Kai platform infrastructure and are used for strategic and SEO quality checks.

kai-growth-hacker