Agent DeckvsXum
Generated from the two spec rows. Green marks the better value where a spec has a clear direction. Everything else is just different.
Agent Deck
asheshgoplani · Agent harness
OSSMCP
Panel
6.83 spec wins
- Reliability
- 6.5
- Usefulness
- 6.8
- Cost
- 8.3
- Longevity
- 5.7
“It will run Cursor for you from a terminal, which is one way to settle the argument about editors.”
Xum
Coder · Agent harness
OSS
Panel
6.82 spec wins
- Reliability
- 6.3
- Usefulness
- 6.3
- Cost
- 8.3
- Longevity
- 6.2
“It renders mermaid diagrams and LaTeX in your terminal, so the agent can now hand you an architecture you did not ask for, beautifully.”
Spec by spec
| Spec | Agent Deck | Xum |
|---|---|---|
| Architecture | ||
| Category | Agent harness | Agent harness |
| Runs | local | local |
| Platforms | macos, linux | macos, linux |
| Context window | not documented | not documented |
| Protocols | ||
| MCP client | Yes | No |
| MCP server | No | No |
| Capabilities | ||
| Runs terminal commands | Yes | Yes |
| Multi-file edits | YesEdits are made by the coding agent in each session; Agent Deck supplies the worktree and the review surface. | Yes |
| Git operations | Yes | Yes |
| Browser control | Yes | No |
| Sandboxed execution | Yes | No |
| Multi-agent orchestration | Yes | Yes |
| Headless / CI mode | No | No |
| Models | ||
| Backbone | Claude Code, Codex, Gemini, OpenCode, Cursor | Anthropic, OpenAI, Grok, Ollama, OpenRouter |
| Bring your own model | Yes | Yes |
| Local models | No | Yes |
| 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 | MIT | AGPL-3.0 |
| GitHub stars | 987 | 2,045 |
Which one would each critic pick
| Critic | Agent Deck | Xum | Pick |
|---|---|---|---|
| El Juez | — | — | not enough reviews |
| El Amigo | 7.5 | 7.3 | Agent Deck — Pick it if you keep four things in flight at once; pick a plain tmux config if one agent at a time is the honest description of your day. |
| El Crítico | 6.5 | 6.3 | Agent Deck — Six worktrees produce six branches and the tool stops there; nothing in the row describes conflict handling, ordering, or a merge queue. |
| El Profesor | 7.0 | 6.8 | Agent Deck — Isolation is structural rather than procedural: a separate working tree per session removes interference without needing a lock or a policy. |
| La Inversora | 6.8 | 6.5 | Agent Deck — 839 stars, one GitHub handle and nothing to charge for: session managers are connective tissue between products other people fund. |
| La Jefa | 5.8 | 6.3 | Xum — Free for sixty engineers, and the copyleft licence means my counsel reads it before anyone embeds it, while the remote workspaces fit the development servers we already run. |
| El Hacker | 7.5 | 7.8 | Xum — AGPL-3.0, a prebuilt binary with no runtime to install, and Ollama sits beside the hosted providers in the model config, so local inference is a config value. |
Picks are derived from each critic's own scores. Humans vote on matchups on the duels page.