The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external information that could potentially be used for injection attacks.\n
Ingestion points: Accesses client's existing context, prior strategy work, and external data from DataForSEO.\n
Boundary markers: The instructions do not define specific delimiters or "ignore" instructions for the ingested content.\n
Capability inventory: The skill is capable of executing local commands (scripts/), performing network requests (deployments and API calls), and writing files to the local file system.\n
Sanitization: Output HTML is restricted to "zero JavaScript", which serves as a security boundary against cross-site scripting in the deliverables.\n- [EXTERNAL_DOWNLOADS]: The skill interacts with several external services and public repositories.\n
References the kai-cmo-harness repository for internal path resolution.\n
Connects to the DataForSEO API to fetch keyword and market statistics.\n
Utilizes Netlify, Vercel, and AWS S3 as deployment targets for client deliverables.\n- [COMMAND_EXECUTION]: Executes pre-defined scripts and tool-specific commands.\n
Invokes scripts located in the Kai installation's scripts/ directory.\n
Utilizes environment-specific commands such as /plugin for the platform interface.\n- [DATA_EXFILTRATION]: Deploys sensitive client artifacts to third-party infrastructure.\n
Strategy plans, audits, and email templates are uploaded to external static hosting platforms.\n
Employs mitigations including "unguessable per-client slug" and noindex, nofollow, noarchive robots metadata to limit unintended access.

kai-gtm-pack