The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external CRM records, including contact notes, messages, and conversation history, which could be used to host instructions targeting the agent.
Ingestion points: The skill retrieves data from HighLevel via execute-operation for entities like contacts, conversations, and form submissions as described in SKILL.md and manifest.yaml.
Boundary markers: No specific delimiters or instructions are provided to the agent to treat retrieved CRM content as untrusted or to disregard potential commands embedded within data fields.
Capability inventory: The skill possesses significant write capabilities, including the ability to send messages, manage appointments, and enroll contacts in workflows via the /opt/hatch/bin/ghl binary.
Sanitization: There is no evidence of content sanitization or validation to filter out potential prompt injection strings from the CRM data before it is presented to the agent.
[COMMAND_EXECUTION]: The skill relies on a local binary located at /opt/hatch/bin/ghl to interact with the HighLevel API.
This execution is limited to the operations defined within the tool's catalog and requires explicit sub-account identification.
Destructive or consequential actions like deletions and messaging are gated by mandatory user approval settings in manifest.yaml.

ghl