api-http-test

Pass

Audited by Gen Agent Trust Hub on Mar 30, 2026

Risk Level: SAFECREDENTIALS_UNSAFEDATA_EXFILTRATIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: The skill collects and stores sensitive authentication data, including passwords, API keys, and bearer tokens, in the .skills/api-http-test/http.toml file. Although the skill attempts to verify that this file is gitignored, the presence of plain-text credentials on the local filesystem represents a security risk.\n- [DATA_EXFILTRATION]: The run_http.sh script implements functionality to read the content of local files using the --body-file parameter and transmit the data to external URLs via HTTP requests. This capability can be leveraged to exfiltrate sensitive files if the agent is manipulated.\n- [COMMAND_EXECUTION]: The skill relies on several shell scripts (api_http_test.sh, bootstrap_profile.sh, infer_auth_mode.sh) and embedded Python scripts to perform its operations, including environment setup and request execution.\n- [EXTERNAL_DOWNLOADS]: The skill makes outbound network connections to user-defined endpoints and authentication servers. This is core functionality but involves interaction with external, potentially untrusted sources.\n- [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes untrusted data from external sources.\n
  • Ingestion points: API response bodies and OpenAPI documentation files (docs/*.yaml).\n
  • Boundary markers: Not utilized in the processing of external data.\n
  • Capability inventory: The agent has the ability to execute shell commands, perform network requests, and read local files.\n
  • Sanitization: No sanitization or validation of the content received from API endpoints or documentation is performed before it is presented to the agent's context.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 30, 2026, 08:15 AM