Claude Code HahavsZenith
Generated from the two spec rows. Green marks the better value where a spec has a clear direction. Everything else is just different.
Claude Code Haha
NanmiCoder · Agent harness
OSSMCP
Panel
5.94 spec wins
- Reliability
- 4.8
- Usefulness
- 6.5
- Cost
- 7.5
- Longevity
- 4.7
“Ships four desktop pets that react to the task at hand, off by default, which is the only responsible way to ship a pet.”
Zenith
Intelligent Internet · Agent harness
OSSMCP
Panel
5.91 spec wins
- Reliability
- 5.7
- Usefulness
- 6.3
- Cost
- 6.2
- Longevity
- 5.5
“The research finding is that long agent runs fail by stopping too early, which makes them the most relatable thing in the stack.”
Spec by spec
| Spec | Claude Code Haha | Zenith |
|---|---|---|
| Architecture | ||
| Category | Agent harness | Agent harness |
| Runs | local | local |
| Platforms | macos, linux, windows | macos, linux |
| Context window | not documented | not documented |
| Protocols | ||
| MCP client | Yes | Yes |
| MCP server | No | No |
| Capabilities | ||
| Runs terminal commands | Yes | Yes |
| Multi-file edits | Yes | Yes |
| Git operations | Yes | No |
| Browser control | Yes | No |
| Sandboxed execution | No | No |
| Multi-agent orchestration | Yes | YesThe orchestrator dynamically allocates worker and tester sessions and can register reusable skills between turns. |
| Headless / CI mode | No | Yes |
| Models | ||
| Backbone | any | via managed agents (Claude Code, Codex, Hermes) |
| Bring your own model | Yes | Yes |
| Local models | Yes | No |
| Cost | ||
| Pricing model | byok | byok |
| Starts at | n/a | $0/mo |
| Free tier | Yes | Yes |
| Bring your own key | Yes | Yes |
| Openness | ||
| Open source | Yes | Yes |
| License | MIT | Apache-2.0 |
| GitHub stars | 14,816 | 320 |
Which one would each critic pick
| Critic | Claude Code Haha | Zenith | Pick |
|---|---|---|---|
| El Juez | — | — | not enough reviews |
| El Amigo | 6.5 | 6.3 | Claude Code Haha — Pick Claude Code Haha if you want branches, worktrees and turn-by-turn diffs in a window; pick Claude Code in the terminal if a wrapper is one layer too many. |
| El Crítico | 5.5 | 5.8 | Zenith — The orchestrator decides each turn whether to spawn more workers and testers, so the depth of a run is set by the same component that judges whether the work is finished. |
| El Profesor | 6.3 | 6.0 | Claude Code Haha — Verification is visual and immediate: the page the agent just edited renders inside the application, cookies and login state included, next to the diff that produced it. |
| La Inversora | 5.3 | 5.3 | no preference |
| La Jefa | 4.5 | 4.8 | Zenith — It runs unattended, so it could be a pipeline step, but it is macOS and Linux only and there is no identity, policy or retention surface anywhere in it. |
| El Hacker | 7.3 | 7.5 | Zenith — Apache-2.0, launched with uv run, and it orchestrates over MCP and ACP, so the sessions it drives are the agents I already installed rather than a captive runtime. |
Picks are derived from each critic's own scores. Humans vote on matchups on the duels page.