fuseraft-cli Documentation¶
fuseraft-cli is a multi-agent orchestration CLI built on Microsoft Agent Framework and Microsoft.Extensions.AI. You define teams of AI agents in a YAML config — each agent has a system prompt, a model, and a set of plugins — and the orchestrator drives them through a conversation until the task is done.
fuseraft-cli is actively maintained and in production use. New features ship regularly.
What it does¶
- Runs any number of agents in a coordinated loop driven by keyword routing, LLM-based selection, or fully autonomous Magentic orchestration
- Gives each agent access to tools: filesystem, shell, git, HTTP, JSON, search, Docker sandboxes, MCP servers
- Saves a checkpoint after every turn so sessions can always be resumed
- Tracks token usage and estimated cost; can enforce a hard spending cap
- Enforces correctness with routing validators that block handoffs unless evidence is present
- Sandboxes agent file and shell access to a configured directory tree
- Applies per-agent execution rings, prompt injection detection, and a hash-chain audit log via the Agent Governance Toolkit
- Supports mixing any combination of LLM providers per agent
- Auto-curates reusable skills from completed sessions and injects relevant ones at session start via a SQLite FTS5 index
- Schedules recurring sessions via cron expressions (
fuseraft schedule add/list/run) - Rotates API keys automatically on 429 rate-limit responses when a key pool is configured
- Accumulates durable cross-session knowledge: architecture decisions, repository graph, provenance claims, repository memory patterns, and long-horizon objectives — all queryable by agents via the adaptive context broker
Guides¶
| Doc | What it covers |
|---|---|
| Getting Started | Prerequisites, installation, first run |
| Writing Effective Tasks | How to write task descriptions that produce correct, verifiable results |
| Spec-Driven Development | Using --spec to anchor agents to an agreed specification before implementation begins |
| CLI Reference | All commands and flags |
| Configuration | Full config schema (YAML and JSON) |
| Models & Providers | Model configuration and auto-detection |
| Plugins | All built-in tools agents can call |
| Strategies | Selection and termination strategies |
| Routing Validators | Anti-hallucination handoff guards |
| Harness Engineering | Building configs that enforce correctness mechanically |
| MCP Integration | Connecting external MCP servers |
| Security & Sandbox | File and network containment |
| Governance | Execution rings, audit log, circuit breaker, SLO tracking |
| Sessions | Resumption, HITL, cost tracking, compaction |
| Context Management | How fuseraft manages context across a long session |
| Context Store | Importing reference material for agents |
| Skills | Portable skill packages, skill curation, and the cross-session skill index |
| Knowledge Layer | ADR registry, repository graph, provenance tracking, repository memory, objectives, context broker, and lifecycle GC |
| Examples | Ready-to-use config examples |
VS Code Extension¶
The Fuseraft VS Code extension brings the CLI into your editor — run tasks, browse sessions, validate configs, and get YAML/JSON IntelliSense, all from the activity bar.