paperclip-create-agent
Pass
Audited by Gen Agent Trust Hub on Jul 6, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
curlcommand-line tool to interact with the Paperclip API. This is the primary method for discovering environment configurations and submitting agent hiring requests. - [DATA_EXFILTRATION]: Network communication is performed with the endpoint specified in the
$PAPERCLIP_API_URLvariable. This transmission of agent metadata and instructions is required for the skill's intended administrative functionality. - [PROMPT_INJECTION]: An indirect prompt injection surface exists as the skill reads and processes existing agent configurations (ingestion points in
SKILL.mdsteps 2 and 3). This risk is mitigated through comprehensive auditing guidelines, including the 'Draft-Review Checklist' (boundary markers) and specialized 'security lenses' (sanitization guidance) that emphasize input validation and least-privilege principles. Capabilities are limited to standardcurlAPI interactions.
Audit Metadata