Overview
An open-source long-horizon SuperAgent harness that researches, codes, and creates. With the help of sandboxes, memories, tools, skill, subagents and message gateway, it handles different levels of tasks that could take minutes to hours.
deer-flow is listed in the db.fyi developer directory under LLM Frameworks. Use official docs and the project repository to verify current capabilities, pricing, and security posture before adopting it in production.
Key capabilities
- Primary focus: An open-source long-horizon SuperAgent harness that researches, codes, and creates. With the help of sandboxes, memories, tools, skill, subagents and message gateway, it handles different levels of tasks that could take minutes to hours.
- Integration surface: see official site for SDKs and APIs
- Pricing model on record: Free (confirm on the vendor site)
Tradeoffs
- Fast-moving open-source and agent tooling can change APIs and defaults without long deprecation windows.
- Claims about autonomy, compression, or “junior developer” behavior should be validated on your own repos and evals.
- Self-hosted options may reduce vendor lock-in but increase ops burden (GPU, upgrades, access control).
Ideal for
- Developers and teams evaluating llm frameworks options in the same cluster as related db.fyi tools
- Prototyping agent or LLM workflows where deer-flow’s stated focus matches the task
- Comparing open alternatives before committing to a paid coding or agent stack
Not ideal for
- Regulated workloads until you verify compliance, data retention, and audit logs from primary sources
- Teams that need a fully managed SLA without reading the project’s support model
- Use cases that require features not documented at https://deerflow.tech
Specs
| Field | Value | |-------|-------| | Category | LLM Frameworks | | Pricing | Free | | Website | https://deerflow.tech | | GitHub | Not publicly documented | | Open source | Yes / see repo | | Last updated | 2026-09-11 |
Comparisons
Browse related tools in the AI directory, task pages, and compare views. Prefer evidence from official docs over third-party roundups.
Last updated
2026-09-11