Open-Source Dynamic Agent Workflow Solution
cohix · reddit · 2026-07-14
The author is building an open-source version of the **Claude Dynamic Workflows** concept: enabling any harness or model to use dynamic workflows and combining multiple agents and models within a single task. ### Design Motivation - Use different models to complement each other's biases and blind spots - Distribute workloads across multiple LLM subscriptions to avoid single-service quota bottlenecks - Leverage both remote and local models simultaneously ### Implementation - Enabled via the `--dynamic` argument in `awman` - Configure an allowed list of agents/models first - Provide "dynamic workflow guidance" to the leader agent - The leader automatically generates a TOML workflow file describing the process - `awman` executes the workflow automatically and mounts a persistent workflow context directory across all agent containers for sharing scripts, notes, and findings - Includes setup/teardown to ensure the work tree, tests, and builds remain intact, automatically launching a repair agent on failure ### Results The author claims this approach works exceptionally well for **large feature specs**, breaking tasks down into 20+ steps with 5-7 agents running in parallel, significantly outperforming manual workflows or single prompts.
Related event: Developers Explore Open-Source Dynamic Agent Workflows(2 posts)→
More from coding & agent
- Hermes Desktop adds multiple instances for agent workflows — Teknium · 2026-07-21
- App Store Rejection: Third-Party AI & HealthKit Data Compliance — JasonBotterill · 2026-07-21
- uv-scripts/ocr returns to the top of Hugging Face datasets with a JSON model picker — vanstriendaniel · 2026-07-21
- Sonar CEO says a guide-verify-solve loop cuts coding-agent issues by 92% — alex_verem · 2026-07-21
- A creator built an Awwwards-style landing page with ChatGPT 5.6 Sol in one conversation — paw_lean · 2026-07-21
- OpenAI’s Build Week buildathon drew 40 people for 11 hours with Codex — paw_lean · 2026-07-21