go-agentvsLegion
Generated from the two spec rows. Green marks the better value where a spec has a clear direction. Everything else is just different.
go-agent
Protocol Lattice · Agent framework
OSS
Panel
6.53 spec wins
- Reliability
- 6.7
- Usefulness
- 5.8
- Cost
- 8.5
- Longevity
- 5.0
“It offers an agent-as-tool pattern, so your agents can now delegate to each other and no single one of them has to be responsible.”
Legion
Software Mansion · Agent framework
OSS
Panel
6.52 spec wins
- Reliability
- 6.0
- Usefulness
- 6.2
- Cost
- 8.0
- Longevity
- 5.8
“It writes Lua to get work done in Elixir, which is two functional languages agreeing that a third one should take the blame.”
Spec by spec
| Spec | go-agent | Legion |
|---|---|---|
| Architecture | ||
| Category | Agent framework | Agent framework |
| Runs | local | local |
| Platforms | macos, linux, windows | macos, linux |
| Context window | not documented | not documented |
| Protocols | ||
| MCP client | No | No |
| MCP server | No | No |
| Capabilities | ||
| Runs terminal commands | No | No |
| Multi-file edits | No | Yes |
| Git operations | No | No |
| Browser control | No | No |
| Sandboxed execution | No | YesLegion runs generated Lua or Elixir in its own sandbox; the README does not describe a container runtime. |
| Multi-agent orchestration | YesSub-agents are exposed with agent-as-tool patterns for specialist agents and hierarchical workflows. | No |
| Headless / CI mode | No | No |
| Models | ||
| Backbone | Gemini, Vertex AI, OpenAI, Anthropic, Ollama, OpenRouter | multiple providers |
| Bring your own model | Yes | Yes |
| Local models | YesOllama is a first-class provider, defaulting to http://localhost:11434. | No |
| Cost | ||
| Pricing model | byok | byok |
| Starts at | $0/mo | $0/mo |
| Free tier | Yes | Yes |
| Bring your own key | Yes | Yes |
| Openness | ||
| Open source | Yes | Yes |
| License | Apache-2.0 | MIT |
| GitHub stars | 258 | 220 |
Which one would each critic pick
| Critic | go-agent | Legion | Pick |
|---|---|---|---|
| El Juez | — | — | not enough reviews |
| El Amigo | 6.8 | 6.8 | no preference |
| El Crítico | 5.8 | 6.0 | Legion — The generated snippet runs in the project's own sandbox, and the README does not describe a container runtime, so the isolation boundary is whatever the host process could already reach. |
| El Profesor | 7.0 | 7.8 | Legion — One evaluation can filter, branch and loop, so a sequence that costs a tool-calling agent a model round trip per step collapses into a single call. The claim is structural and checkable. |
| La Inversora | 5.8 | 6.3 | Legion — 166 stars and a consultancy's name on it. Software Mansion builds open libraries to sell engineering time, which funds this exactly as long as the pitch still works. |
| La Jefa | 6.0 | 5.5 | go-agent — The one governance feature I care about is here: token-budget middleware, so spend is a limit in code rather than a surprise on an invoice. |
| El Hacker | 7.8 | 6.8 | go-agent — Apache-2.0, Ollama defaults to localhost:11434, and instructions load from a .skills directory in the working directory rather than a vendor's cloud. |
Picks are derived from each critic's own scores. Humans vote on matchups on the duels page.