99vsRefact.ai
Generated from the two spec rows. Green marks the better value where a spec has a clear direction. Everything else is just different.
99
ThePrimeagen · IDE extension
#219OSS
Panel
5.61 spec wins
- Reliability
- 4.8
- Usefulness
- 5.5
- Cost
- 7.5
- Longevity
- 4.7
“It ships for macOS and Linux only, which is one way to make sure Windows developers keep hand-coding.”
Refact.ai
Refact.ai (Small Magellanic Cloud AI) · IDE extension
#192OSSMCP
Panel
5.37 spec wins
- Reliability
- 5.0
- Usefulness
- 5.3
- Cost
- 7.3
- Longevity
- 3.3
“Ships eight agent modes including one called buddy, for when the other seven feel too professional.”
Spec by spec
| Spec | 99 | Refact.ai |
|---|---|---|
| Architecture | ||
| Category | IDE extension | IDE extension |
| Runs | local | local |
| Platforms | macos, linux | macos, linux, windows |
| Context window | not documented | not documented |
| Protocols | ||
| MCP client | No | Yes |
| MCP server | No | No |
| Capabilities | ||
| Runs terminal commands | No | Yes |
| Multi-file edits | Yes | Yes |
| Git operations | No | Yes |
| Browser control | No | Yes |
| Sandboxed execution | No | No |
| Multi-agent orchestration | No | Yes |
| Headless / CI mode | No | Yes |
| Models | ||
| Backbone | Claude Sonnet via OpenCode, Claude Code | Claude, GPT, Gemini, OpenRouter, Ollama |
| Bring your own model | YesModel choice is delegated to the configured provider CLI; if no model is set the provider's default is used. | Yes |
| Local models | No | Yes |
| Cost | ||
| Pricing model | byok | byok |
| Starts at | n/a | n/a |
| Free tier | Yes | Yes |
| Bring your own key | Yes | Yes |
| Openness | ||
| Open source | Yes | Yes |
| License | unspecified | BSD-3-Clause |
| GitHub stars | 4,744 | 69 |
Which one would each critic pick
| Critic | 99 | Refact.ai | Pick |
|---|---|---|---|
| El Juez | — | — | not enough reviews |
| El Amigo | 6.3 | 5.5 | 99 — Pick it if you hand-code in Neovim and want an agent that assists rather than takes over; pick CodeCompanion.nvim if you want the fuller editor experience. |
| El Crítico | 5.5 | 5.3 | 99 — It delegates every hard part to a provider CLI it does not control, and the README calls the project beta with an API that still changes. |
| El Profesor | 5.5 | 6.0 | Refact.ai — One Rust binary supervises a warm worker per project, gathers context from AST and semantic search over a local vector index, and applies edits through apply_patch; principled. |
| La Inversora | 5.0 | 3.0 | 99 — 4,752 stars attached to a personal project by a well-known creator, with no company, no price and no revenue line behind it. |
| La Jefa | 4.5 | 4.0 | 99 — Zero seat cost, and zero of everything else procurement asks for: no vendor, no support, no audit trail, and the bill sits with whichever agent CLI it drives. |
| El Hacker | 7.0 | 7.8 | Refact.ai — BSD-3-Clause, providers in ~/.config/refact/ including Ollama, LM Studio, vLLM and even GitHub Copilot as a backend, project state in .refact/, MCP documented in the wiki; entirely mine. |
Picks are derived from each critic's own scores. Humans vote on matchups on the duels page.