together-webhooks-events

Pass

Audited by Gen Agent Trust Hub on Sep 10, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCREDENTIALS_UNSAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests job output and error artifacts from Together AI via WebFetch, creating a potential surface for indirect prompt injection if these artifacts contain malicious instructions.
  • Ingestion points: Job output and error files fetched during the job completion and reconciliation phase (SKILL.md).
  • Boundary markers: Instructions advise mapping data to a versioned internal event schema, which provides structure, but lacks explicit boundary delimiters for potentially unstructured data in the artifacts.
  • Capability inventory: The skill has access to WebFetch for data retrieval and Write/Edit for event persistence.
  • Sanitization: The skill mentions schema mapping but does not specify validation or sanitization requirements for the content retrieved from Together AI.
  • [EXTERNAL_DOWNLOADS]: The skill references and fetches information from Together AI official documentation and SDK repositories (docs.together.ai, github.com/togethercomputer). These are recognized well-known service resources.
  • [CREDENTIALS_UNSAFE]: Mentions the use of TOGETHER_API_KEY. The skill correctly includes instructions to maintain credential confidentiality, advising against exposing the key to callback consumers and recommending the use of approved secret stores.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 10, 2026, 06:11 PM
Security Audit — agent-trust-hub — together-webhooks-events