agent-id-proxy

Pass

Audited by Gen Agent Trust Hub on Aug 28, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: Executes the local proxy CLI using the command node *agent-id-proxy/bin/cli.mjs:* to initialize and manage the credential-injecting proxy service.
  • [COMMAND_EXECUTION]: Uses Bash(curl:*) to communicate with the local proxy and perform external API calls, granting the agent the ability to execute any curl command.
  • [DATA_EXFILTRATION]: The broad curl authorization allows the agent to send data to any external URL, creating a potential channel for exfiltrating sensitive information.
  • [DATA_EXFILTRATION]: Accesses local sensitive files such as ~/.agent-id-pass (vault passphrase) and ~/.agent-id-proxy-token (proxy auth token) to manage secure sessions.
  • [INDIRECT_PROMPT_INJECTION]: The agent ingests data from external APIs via the proxy, which could contain malicious instructions if the upstream source is compromised.
  • Ingestion points: API responses received through http://localhost:48771/ (documented in SKILL.md).
  • Boundary markers: No boundary markers or 'ignore' instructions are present to prevent the agent from following instructions embedded in API responses.
  • Capability inventory: The skill possesses the ability to execute shell commands (node, curl) and read files.
  • Sanitization: The proxy forwards upstream data verbatim without sanitization or validation of the content for prompt injection patterns.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 28, 2026, 04:21 PM
Security Audit — agent-trust-hub — agent-id-proxy