The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes Node.js scripts (scripts/search.mjs and scripts/request.mjs) to perform CRM data lookups. The agent is instructed to execute these scripts with various actions and options based on user queries.
[DATA_EXFILTRATION]: The skill performs network requests to an external domain (crm.langcore.net) to retrieve CRM data. While this is the intended functionality of the tool, it involves sending user-provided query parameters and an authentication token to a non-whitelisted external service.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted data from the CRM API (e.g., customer descriptions, business records, or lead notes).
Ingestion points: Data enters the agent's context through API responses fetched in scripts/search.mjs and scripts/request.mjs.
Boundary markers: The skill does not implement specific delimiters or warnings to treat fetched CRM content as untrusted data.
Capability inventory: The skill has the ability to execute shell commands via Node.js scripts and make further network requests.
Sanitization: There is no evidence of sanitization or filtering of the content retrieved from the CRM before it is presented to the agent.
[CREDENTIALS_UNSAFE]: The skill requires the LCRM_API_KEY environment variable for authentication. While it does not hardcode secrets, it handles sensitive bearer tokens in the Authorization header during API requests in scripts/lib.mjs.

lcrm-search