agentboards.org
Compare/Efrit vs Roo Code

EfritvsRoo Code

Generated from the two spec rows. Green marks the better value where a spec has a clear direction. Everything else is just different.

Efrit
Steve Yegge · IDE extension
#259OSS
Panel
5.5
0 spec wins
Reliability
4.7
Usefulness
5.8
Cost
7.3
Longevity
4.3

“Three entry points named efrit-chat, efrit-do and efrit-agent, because Emacs was never going to settle for one.”

Roo Code
Roo Code, Inc. (now Roomote) · IDE extension
#185OSSMCP
Panel
4.1
5 spec wins
Reliability
4.3
Usefulness
4.7
Cost
5.5
Longevity
2.0

“Shut down in May and archived, but the extension still runs, which is more than the paid Cloud tier can say.”

Spec by spec

SpecEfritRoo Code
Architecture
CategoryIDE extensionIDE extension
Runslocallocal
Platformsmacos, linux, windowsmacos, linux, windows
Context windownot documentednot documented
Protocols
MCP clientNoYes
MCP serverNoNo
Capabilities
Runs terminal commandsYesBoth shell_exec and eval_sexp are available, and the README warns that AI-generated code executes directly in your Emacs. Yes
Multi-file editsYesYes
Git operationsYesRead-only version-control tools — status, diff, log and blame — are exposed to the model. Yes
Browser controlNoNo
Sandboxed executionNoNo
Multi-agent orchestrationNoYes
Headless / CI modeNoNo
Models
BackboneClaudeAnthropic, OpenAI, Google, OpenRouter, Ollama, LM Studio
Bring your own modelNoYes
Local modelsNoYes
Cost
Pricing modelbyokbyok
Starts at$0/mo$0/mo
Free tierYesYes
Bring your own keyYesYes
Openness
Open sourceYesYes
LicenseunspecifiedApache-2.0
GitHub stars45524,287

Which one would each critic pick

CriticEfritRoo CodePick
El Juez——not enough reviews
El Amigo6.34.5Efrit — Pick it if Emacs is where you actually work and you want the agent inside it; pick Aider in a terminal if you are only visiting Emacs for the keybindings.
El Crítico5.33.8Efrit — The stated design gives the model shell execution and elisp evaluation with nothing on the client deciding what is safe, and the README flags that itself.
El Profesor6.04.3Efrit — The version-control tools exposed to the model are read-only, status and diff and log and blame, which is a deliberate asymmetry between what may be inspected and what may be changed.
La Inversora5.82.5Efrit — One author, 454 stars, no company, and a hard binding to a single model vendor: the survival question here is not funding, it is whether one API keeps its shape.
La Jefa4.82.5Efrit — Nothing per seat and nothing to roll out: the sixty engineers who would use this are the four who already run Emacs, and no record survives of what any of them did.
El Hacker5.37.3Roo Code — Archived, but Apache-2.0, so the fork lives; ZooCode is where I would point my config, and the config barely changes.

Picks are derived from each critic's own scores. Humans vote on matchups on the duels page.

Want a third column? The compare tool handles any two agents. Three-way comparisons are on the roadmap once the spec rows are all verified.