Agent / Orchestration for AI coding
Every agent / orchestration option Flowpicker tracks, with pricing, setup notes, and compatibility for each.
- None — Projects that only need inline completions and chat
- Aider (architect) — Plan-then-edit workflow with git safety net for complex changes
- SWE-agent — Autonomous issue resolution and PR generation from GitHub issues
- Claude Code — Anthropic CLI agent with full repo awareness and terminal access
- Custom LangGraph — Build custom multi-agent workflows with full control and state management
- Cline Agent — Fully autonomous agentic coding in VS Code with Plan/Act modes and MCP
- Devin — Hands-off task completion with a managed cloud agent that works asynchronously
- Windsurf Cascade Agent — Hybrid local and cloud agents with unified command center and MCP ecosystem
- OpenHands Agent — Open-source autonomous coding agent with sandboxed execution and multi-agent support
- OpenAI Codex CLI Agent — Official OpenAI CLI agent for sandboxed coding tasks with multi-model support
- Sourcegraph Cody Agent — Cross-repo AI coding assistance with Sourcegraph-powered context awareness
- Tabnine Agent — Enterprise agentic coding with organizational context, coaching, and full privacy controls
- AutoGen — Multi-agent conversations, complex research workflows, human-in-the-loop automation
- CrewAI — Team-based workflows, content creation pipelines, hierarchical task execution
- Pydantic AI — Production agents that need validated structured outputs and type-safe tool calls
- Smolagents — Code-generation-first agents, Hugging Face model ecosystem, research use cases
- OpenAI Codex (Cloud) — Background async tasks (issue triage, refactors, doc generation) running in a managed cloud sandbox
- Jules (Google) — GitHub issue-driven async coding with managed cloud sandbox and PR output
- Factory Droid — Enterprise teams wanting multi-droid orchestration across repos, tickets, and incidents
- Cursor Background Agent — Spinning up many parallel agents from Cursor that run async in the cloud and return PRs
- Claude Code SDK Agent — Embedding the Claude Code agent loop into custom apps, CI, or background workers
- Mastra — TypeScript developers building multi-step agents with workflows, RAG, evals, and deploys
- Agno (Phidata) — Lightweight Python multi-agent systems with built-in storage, memory, and tools
- LangChain (Agent Core) — Composable chains, RAG, and tool-using agents with deep ecosystem integrations
- Semantic Kernel — Enterprise .NET / Python teams orchestrating LLMs with plugins, planners, and memory
- OpenDevin (legacy) — Reference implementation of the original Devin-style agent loop
- Bolt.new (StackBlitz) — Prompt-to-app generation in the browser, instant deploy, full-stack JS/TS scaffolds
- Lovable — Non-developers and founders building MVPs with React + Supabase from natural-language prompts
- Replit Agent 3 — In-browser autonomous app generation with built-in deploy, DB, and test agents
- All Hands Cloud — Teams wanting OpenHands without managing their own runtime; GitHub-native triggers