The Agent Skills Directory

[PROMPT_INJECTION]: Indirect prompt injection surface identified.
Ingestion points: Reads criteria and ICP data from claude-code-gtm/context/{company}_context.md and claude-code-gtm/context/{vertical-slug}/hypothesis_set.md.
Boundary markers: No delimiters or isolation instructions are present to prevent content in these files from being interpreted as agent commands.
Capability inventory: Performs network operations to api.extruct.ai and basic shell validation. No high-privilege system or file-write capabilities are used.
Sanitization: Content from context files is interpolated directly into API query strings and Discovery criteria without filtering or validation.
[EXTERNAL_DOWNLOADS]: Fetches updated API documentation and endpoint specifications from the vendor's official site at extruct.ai. This is used for schema verification and local reference updates.
[COMMAND_EXECUTION]: Uses test and echo shell built-ins to verify that the EXTRUCT_API_TOKEN is set in the environment before making network requests.

list-building