fishjam-platform

Fail

Audited by Gen Agent Trust Hub on Nov 3, 2026

Risk Level: CRITICALEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill references official SDKs and documentation hosted on github.com/fishjam-cloud/* and the fishjam.io / fishjam.swmansion.com domains. These represent standard vendor resources. Automated scanner alerts regarding fishjam.io were reviewed and identified as false positives for the official product domain.
  • [INDIRECT_PROMPT_INJECTION]: The Fishjam platform allows users to set arbitrary metadata on peers and tracks, which is then broadcast to other participants. While this is a standard feature of the communication API, it represents a potential surface for indirect prompt injection if the metadata is processed without sanitization by an agent.
  • Ingestion points: metadata fields in createPeer and update_peer_metadata calls (auth-model.md, glossary.md).
  • Boundary markers: The documentation does not specify the use of delimiters or sanitization instructions for these strings.
  • Capability inventory: The skill documents extensive API capabilities for room and peer management (rest-endpoints.md).
  • Sanitization: Metadata is treated as an opaque string by the documented platform.
Recommendations
  • Contains 9 malicious URL(s) - DO NOT USE
Audit Metadata
Risk Level
CRITICAL
Analyzed
Nov 3, 2026, 01:17 AM