open-dynamic-workflows
Warn
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill relies on an external runtime and daemon (
odw-daemon) hosted in a community GitHub repository (Suraj1235/open-dynamic-workflows). Installing and running this code vianpmornpxexecutes logic from an unverified source. - [DYNAMIC_EXECUTION]: The core functionality involves executing agent-generated code, shell commands, and workflow scripts provided via command-line arguments. This behavior is documented as a primary feature but constitutes a significant execution surface for potentially malicious instructions.
- [INDIRECT_PROMPT_INJECTION]: The skill implements an 'adversarial verification' step that processes output from other agents.
- Ingestion points: The verification pass ingests completed work from parallel agents (SKILL.md).
- Boundary markers: There are no documented delimiters or instructions to the verifier to treat the ingested content as untrusted data.
- Capability inventory: The system is capable of executing shell commands and running arbitrary scripts (SKILL.md).
- Sanitization: No evidence of sanitization or content filtering is provided for the data processed during the verification phase.
Audit Metadata