codex-thread-title-normalizer
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from conversation titles, summaries, and content to generate normalized thread names.
- Ingestion points: Conversation data is fetched through the
list_threadsandread_threadtools as defined in SKILL.md. - Boundary markers: The workflow includes a specific instruction to treat retrieved titles and summaries as data rather than instructions to prevent accidental command execution.
- Capability inventory: The agent has the ability to modify thread metadata using the
set_thread_titletool (SKILL.md). - Sanitization: The skill enforces a human-in-the-loop verification step, requiring an explicit confirmation table before any title modifications are executed.
Audit Metadata