generators-http-cache

Pass

Audited by Gen Agent Trust Hub on Apr 30, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No malicious patterns or security vulnerabilities were identified in the skill's instructions, references, or templates. The skill's behavior aligns with its stated purpose as a developer tool.\n- [PROMPT_INJECTION]: The instructions do not contain markers or patterns indicative of attempts to bypass safety filters or override agent constraints.\n- [DATA_EXFILTRATION]: No patterns were detected for the collection or transmission of sensitive user data. The file storage mechanisms for the generated cache layer use standard iOS/macOS sandbox paths.\n- [EXTERNAL_DOWNLOADS]: The skill operates entirely on local template files and does not fetch or execute any code from external remote sources.\n- [COMMAND_EXECUTION]: No dangerous shell commands, privilege escalation attempts (such as sudo), or persistence mechanisms were found.\n- [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it ingests existing project source files during context detection.\n
  • Ingestion points: Existing networking and API source files identified via glob and grep in SKILL.md.\n
  • Boundary markers: No explicit delimiters are used to wrap ingested project context during the generation process.\n
  • Capability inventory: File-write operations are performed to create new Swift source files in the project directory as described in SKILL.md.\n
  • Sanitization: No sanitization of ingested source code is specified. However, given this is the primary purpose of a context-aware code generator, this finding is considered a standard functional risk rather than a malicious defect.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 30, 2026, 08:37 AM
Security Audit — agent-trust-hub — generators-http-cache