agents-swarm-orchestration

Fail

Audited by Gen Agent Trust Hub on Oct 17, 2026

Risk Level: CRITICALDYNAMIC_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The test script located at scripts/test_wave_recipe.py implements logic to parse markdown files (references/recipe-wave-dispatch.md and references/scripted-workflows.md) and execute the extracted code blocks. It uses subprocess.run to invoke bash and node -e on content extracted at runtime, creating a risk of arbitrary code execution if the documentation files are modified.\n- [COMMAND_EXECUTION]: The 'Wave Dispatch' recipe provided in references/recipe-wave-dispatch.md provides shell scripts (orchestrator.sh, worker.sh, synthesizer.sh) that dynamically construct and execute commands via the claude CLI. This orchestration pattern involves piping task payloads and file paths into shell commands, which presents an attack surface for command injection.\n- [EXTERNAL_DOWNLOADS]: The skill's data/sources.json file includes references to numerous third-party GitHub repositories (e.g., smtg-ai/claude-squad, craigsc/cmux, byt3bl33d3r/figaro) and community blogs. These resources are provided as orchestration examples but originate from unvetted external sources outside the established trusted vendor list.\n- [INDIRECT_PROMPT_INJECTION]: The orchestration model is inherently susceptible to indirect prompt injection as it facilitates the ingestion of data from subagents, MCP servers, and external documentation. Although the skill includes dedicated 'Operational Guardrails' and 'Worker Self-Rejection' patterns to mitigate these risks, the high degree of automation and tool-chaining creates a significant exposure surface.\n- [EXTERNAL_DOWNLOADS]: Fetches official documentation and SDK examples from OpenAI's GitHub pages (openai.github.io). These references were flagged by automated scanners but originate from a trusted organization.
Recommendations
  • Contains 4 malicious URL(s) - DO NOT USE
Audit Metadata
Risk Level
CRITICAL
Analyzed
Oct 17, 2026, 08:31 AM