The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill directs the agent to install the gohighlevel-pp-cli binary using npx from the @mvanhorn/printing-press-library npm package or via go install from the author's GitHub repository. These are external dependencies required for the skill's functionality.
[DATA_EXFILTRATION]: The CLI tool provides a --deliver webhook:<url> feature, which allows the agent to transmit command results (potentially containing sensitive CRM data like contacts, opportunities, and messages) to any external URL. Furthermore, the feedback command can be configured to send local data to a remote endpoint via the GOHIGHLEVEL_FEEDBACK_ENDPOINT environment variable.
[COMMAND_EXECUTION]: The skill's primary operation involves executing shell commands to interact with the gohighlevel-pp-cli tool. This includes performing bulk operations and executing arbitrary read-only SQL queries against a local SQLite mirror of CRM data.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests and processes untrusted data from the GoHighLevel API and local database mirrors (such as contact notes and message threads).
Ingestion points: Data retrieved via contacts search, convo thread, and sql query results.
Boundary markers: The instructions lack explicit delimiters or warnings to the agent to ignore instructions embedded within the CRM data.
Capability inventory: The agent has the ability to execute shell commands and perform network requests via the CLI's webhook feature.
Sanitization: There is no evidence of sanitization or filtering applied to the data retrieved from the CRM before it is processed by the agent.

pp-gohighlevel