Claudio

Claudio is a terminal-native coding-agent harness implemented primarily as a single Go binary. Its current repository documents multi-agent teams with delegation and messaging, persistent memory, a Vim-style terminal UI, session crystallization, MCP and LSP integration, scheduled agent tasks, a browser/PWA command center and multi-provider model routing. ¹

The repository also documents local or hosted model routing through Anthropic, OpenAI-compatible providers, Groq, Ollama, Together and vLLM, plus one-shot and resumable terminal workflows. The repository surface shows substantial development history, with 512 commits surfaced at review time. These are first-party/project facts, not AiToolMap hands-on findings. ¹

The repository is publicly downloadable and documents installation through `go install`. AiToolMap found no separate paid Claudio software tier in the verified surface. Hosted model or provider use can still create separate third-party subscription or API costs. ¹

The README describes Claudio as open-source, but the indexed main repository surface did not expose a verifiable license file. A separate plugin, claudio-codex, explicitly uses MIT, but that license is not transferred to the main Claudio repository. AiToolMap therefore records main-product licensing as unverified rather than assuming MIT. ^{1 2}

The current repository is detailed and active-looking, but the canonical 50-source panel found no exact independent review, benchmark, security assessment or eligible numeric rating input for Claudio. That means capability, efficiency and safety assertions remain primarily first-party until independently reproduced. No hands-on testing was performed.

The license ambiguity is also material for organizations that need clear redistribution or modification rights: public source availability alone does not establish an open-source license.

Claudio fits experienced developers who prefer a terminal-native environment and want several agentic-development primitives in one stack rather than stitching together separate tools for orchestration, memory, MCP/LSP, scheduled jobs and remote session management. ¹

It is less suitable for teams that require independently benchmarked reliability, enterprise support commitments or immediately verifiable licensing terms.