client-relationship-manager
Pass
Audited by Gen Agent Trust Hub on Mar 30, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [PROMPT_INJECTION]: The skill contains an indirect prompt injection surface by design. It instructs the agent to read and process content from text files in the
crm/clients/directory that may contain data from external sources (e.g., email content or meeting notes). - Ingestion points: Client record files (
crm/clients/[client-slug].md), pipeline tracking (crm/pipeline.md), and follow-up lists (crm/follow-ups.md). - Boundary markers: Absent. The instructions do not specify delimiters or instructions to ignore embedded commands within the CRM files.
- Capability inventory: The agent can read files, create new client files, and update records based on user input.
- Sanitization: None. The skill relies on the agent's default processing of markdown text files.
- [COMMAND_EXECUTION]: The skill provides a shell command example for the
openclawCLI to schedule automated CRM reviews. This command sets up a persistent cron task to analyze CRM data at regular intervals. - [DATA_EXFILTRATION]: The optional automated review feature (via the
openclawcron job) is designed to transmit summarized pipeline and client data to an external Telegram chat ID specified by the user.
Audit Metadata