Skills
skills/agentera/agently-skills/agently-triggerflow-interrupts-and-stream/Gen Agent Trust Hub

agently-triggerflow-interrupts-and-stream

Pass

Audited by Gen Agent Trust Hub on Mar 11, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill's implementation patterns for handling runtime streams present a surface for potential indirect prompt injection. \n
  • Ingestion points: Untrusted data enters the workflow context via data.value in the stream_reply pattern in references/model-stream-bridge.md. \n
  • Boundary markers: The documentation lacks examples of using delimiters or instructions to the agent to ignore instructions embedded within the processed data chunks. \n
  • Capability inventory: The skill utilizes async_put_into_stream and async_stop_stream within TriggerFlowRuntimeData to manage live output. \n
  • Sanitization: There is no evidence of data sanitization, validation, or escaping before data is placed into the runtime stream in the provided code examples.\n- [EXTERNAL_DOWNLOADS]: The skill references source code and documentation from the author's official repositories.\n
  • Evidence: Links to the Agently GitHub repository (github.com/AgentEra/Agently) and official documentation (agently.tech) are provided in references/source-map.md to assist with implementation.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 11, 2026, 05:52 PM