Skip to content
ADE Index

Crab Code vs Hermes Agent vs Codex CLI vs OpenCode

Up to four tools, row by row. We do not guess: a cell we have not checked says so.

This page compares the four CLI agents only. Codex has 5 other listings, each with its own page.

How they differ

Pick this if…

Paraphrased from each maker’s own description. Select a line for the quote.

Checked for Hermes Agent, Codex CLI and OpenCode, not yet for Crab Code: Price (Hermes Agent: ; Codex CLI: ; OpenCode: ), Bring your own key (Hermes Agent: ; Codex CLI: ; OpenCode: )

Checked for Crab Code, Codex CLI and OpenCode, not yet for Hermes Agent: Systems (Crab Code: ; Codex CLI: , , , ; OpenCode: , )

Not yet checked for any of these tools: agents it runs.

Dotted values open their quote.

What each can do

Select an answer to see the maker’s sentence, or what we searched.

We have not checked Crab Code’s capabilities yet. 29 capabilities are checked for at least two of these tools. One gets a different answer: Editor integration; 9 differ only in a limit. 5 others are checked for one tool only.

Differences only10
  • Yesthe maker documents it
  • With add-onAdd-onsomething the maker publishes that you install or run
  • *its limit or add-on is named under the row
  • Emptynot yet checked
  • From the listingthe listing’s own facts; not one of the capabilities we check

Models, keys and usage

Named model providersFrom the listing
,,
,,,,,,,,
+30 more model providers
,,,,,,,,,,,,,,,,,,,,,,,,,,,,,
Listing also names 11 othersSpaceXSI (Grok), Amazon Bedrock, Moonshot AI, Vercel AI Gateway, OpenCode Console (OpenCode Zen), Alibaba (Qwen), Meta AI (Llama), Microsoft Foundry (Azure AI Foundry), Nebius Token Factory (Nebius AI Studio), Novita AI, Command Code
,,,
Listing also names: LM Studio
,,,,,,,,
+2 more model providers
,
and others; list may be partial
Listing also names: SpaceXSI (Grok), GitHub Copilot
Local modelsNot yet checked: Crab Code
Not yet checked for Crab Code
Custom OpenAI-compatible endpointNot yet checked: Crab Code
Not yet checked for Crab Code
Token and cost usageNot yet checked: Crab Code
Not yet checked for Crab Code

MCP, hooks, skills and subagents

MCP serversNot yet checked: Crab Code
Not yet checked for Crab Code
local and remote
local and remote
local and remote
HooksNot yet checked: Crab Code
Not yet checked for Crab Code
SkillsNot yet checked: Crab Code
Not yet checked for Crab Code
SubagentsNot yet checked: Crab Code
Not yet checked for Crab Code
PluginsNot yet checked: Crab Code
Not yet checked for Crab Code
Custom commandsNot yet checked: Crab Code
Not yet checked for Crab Code

Editors, scripts and CI

Editor integrationNot yet checked: Crab Code
Not yet checked for Crab Code
ACP editors
Codex editor extension
ACP clients such as Zed
Headless modeNot yet checked: Crab Code
Not yet checked for Crab Code
JSON outputNot yet checked: Crab Code
Not yet checked for Crab Code
GitHub integrationNot yet checked: Crab Code, Hermes Agent, OpenCode
Not yet checked for Crab Code
Not yet checked for Hermes Agent
Codex GitHub Action
Not yet checked for OpenCode

Review, steer and undo

Plan modeNot yet checked: Crab Code, Hermes Agent
Not yet checked for Crab Code
Not yet checked for Hermes Agent
Read-only modeNot yet checked: Crab Code, Hermes Agent
Not yet checked for Crab Code
Not yet checked for Hermes Agent
Plan profile; shell still asks
Review diffsNot yet checked: Crab Code
Not yet checked for Crab Code
Needs checkpoints on
In ACP permission requests
Undo file editsNot yet checked: Crab Code, Codex CLI
Not yet checked for Crab Code
Opt-in; local and SSH only
Not yet checked for Codex CLI

Parallel agents

Parallel sessionsNot yet checked: Crab Code, Codex CLI
Not yet checked for Crab Code
Not yet checked for Codex CLI
Git worktreesNot yet checked: Crab Code
Not yet checked for Crab Code
Local execution only

Context and memory

Project instruction filesNot yet checked: Crab Code
Not yet checked for Crab Code
Memory across sessionsNot yet checked: Crab Code, OpenCode
Not yet checked for Crab Code
Off by default; local memories
Not yet checked for OpenCode
Resume a sessionNot yet checked: Crab Code
Not yet checked for Crab Code
Browse past sessionsNot yet checked: Crab Code
Not yet checked for Crab Code

Remote and mobile control

Remote controlNot yet checked: Crab Code, OpenCode
Not yet checked for Crab Code
To a running app server
Not yet checked for OpenCode
Steer from a phoneNot yet checked: Crab Code, Codex CLI, OpenCode
Not yet checked for Crab Code
Unblocks Kanban workers
Not yet checked for Codex CLI
Not yet checked for OpenCode
NotificationsNot yet checked: Crab Code
Not yet checked for Crab Code

Sandbox, permissions and privacy

OS-level sandboxNot yet checked: Crab Code, Hermes Agent, OpenCode
Not yet checked for Crab Code
Not yet checked for Hermes Agent
Linux/WSL2 needs bubblewrap
Not yet checked for OpenCode
Approval modesNot yet checked: Crab Code
Not yet checked for Crab Code
Runs in a containerNot yet checked: Crab Code, Codex CLI
Not yet checked for Crab Code
Official Docker setup
Not yet checked for Codex CLI
Official Docker image

Images, web and your computer

Image inputNot yet checked: Crab Code
Not yet checked for Crab Code
Needs an image-capable model
Fetch URLsNot yet checked: Crab Code, Hermes Agent, Codex CLI
Not yet checked for Crab Code
Not yet checked for Hermes Agent
Not yet checked for Codex CLI
Browser automationNot yet checked: Crab Code, Codex CLI, OpenCode
Not yet checked for Crab Code
Not yet checked for Codex CLI
Not yet checked for OpenCode
Voice inputNot yet checked: Crab Code, OpenCode
Not yet checked for Crab Code
Only in builds with voice
Not yet checked for OpenCode
Not yet checked: , , ,

3 capabilities are not yet checked for any of these tools and are not shown. Each answer is about the CLI agent in its column. Something the maker publishes separately, such as an editor plugin or a GitHub Action, shows as “With add-on”, even when it has its own listing here. How capabilities are checked

About these tools

Crab Code

Hermes Agent

Codex CLI

OpenCode

How to read the Index score

At a glance

Index score out of 100

Crab CodeNot enough data yetHermes Agent48Codex CLI91OpenCode79

Mindshare HN mentions, 90 days

Crab Code0Hermes Agent84Codex CLI4,237OpenCode1,414

Adoption GitHub stars, not scored

Crab Code77Hermes Agent264kCodex CLI136kOpenCode228k

Release days last 90 days

Crab Code0Hermes Agent19Codex CLI32OpenCode28

Lowest monthly plan billed monthly, lower is cheaper

Crab CodeNot yet checkedHermes AgentNot yet checkedCodex CLI$20/monthOpenCode$10/month

Platforms systems and surfaces

Crab Code2Hermes AgentNot yet checkedCodex CLI4OpenCode3

Votes can only raise a score: up to 20% of it reflects upvotes, graded on a curve against the most-voted tool. The square-root curve has a floor of 100 effective votes. Votes decay with a 90-day half-life. Maker self-votes do not count. Counts below 10 are hidden for display only. After activation, the score is 80% mindshare plus 20% of the greater of mindshare and the vote score. GitHub stars do not affect the Index score. Mindshare uses a log scale; votes use the square-root curve and a 100-vote floor. Ranks use unrounded scores. Mindshare and adoption bars are log-scaled; the others are linear within this comparison. How the score is built

Differences only
4 of 4
Crab Codelingcoder · CLI agent
Hermes AgentCLI agent
Codex CLIOpenAI · CLI agent
OpenCodeAnomaly · CLI agent
Interface
Codex CLI explaining a codebase, with an updated plan checklist
Terminal UI with an inline diff, passing bun tests and a context sidebar
What it is
Open-source alternative to Claude Code, built from scratch in Rust.
Terminal agent for writing, reviewing and running code in local projects
Coding agent from OpenAI that runs locally on your computer
Coding agent with switchable build and read-only plan agents
Category
CLI agent
CLI agent
CLI agentCodex family
CLI agent
Pricing
Starts at (billing shown)
Not yet checked
Lowest monthly plan
Not yet checked
Not yet checked
Free tier
Not yet checked
Not yet checked
Plans
Not yet checked
Bring your own key
Not yet checked
Yes
Yes
Yes
Runs on
Linux
Not yet checked
Not yet checked
Yes
Not yet checked
Terminal
Yes
Not yet checked
Yes
Yes
Web
Not yet checked
Not yet checked
Not yet checked
Yes
Windows
Yes
Not yet checked
Yes
Yes
macOS
Not yet checked
Not yet checked
Yes
Not yet checked
Models and agents
Models
Not yet checked
Not yet checked
GPT-6.2 AstraGPT-6.2 SolGPT-6.1 TerraGPT-6.1 SolGPT-6 Luna
Not yet checked
Drives these agents
Not yet checked
Not yet checked
Not yet checked
Not yet checked
Source and activity
Source
GitHub stars
Most
Release days, last 90 days
0
19
32Most
28
Release days per week
Latest release, any channel
Not yet checked
v2.3.2 18 Nov 2026
Index score
Score and rank
Not enough data yet
48 #14 Score 48.2; ranked on the unrounded score
91 #2 Score 90.7; ranked on the unrounded score
79 #3 Score 78.8; ranked on the unrounded score
Mindshare
00 Hacker News mentions in the last 90 days
4884 Hacker News mentions in the last 90 days
914,237 Hacker News mentions in the last 90 days
791,414 Hacker News mentions in the last 90 days
Adoption, not scored
3677 stars, read 18 Nov 2026
100264,230 stars, read 18 Nov 2026
97135,675 stars, read 18 Nov 2026
100227,631 stars, read 18 Nov 2026
Votes
—Votes can only raise a score: up to 20% of it reflects upvotes, graded on a curve against the most-voted tool. The square-root curve has a floor of 100 effective votes. Votes decay with a 90-day half-life. Maker self-votes do not count. Counts below 10 are hidden for display only.
—Votes can only raise a score: up to 20% of it reflects upvotes, graded on a curve against the most-voted tool. The square-root curve has a floor of 100 effective votes. Votes decay with a 90-day half-life. Maker self-votes do not count. Counts below 10 are hidden for display only.
—Votes can only raise a score: up to 20% of it reflects upvotes, graded on a curve against the most-voted tool. The square-root curve has a floor of 100 effective votes. Votes decay with a 90-day half-life. Maker self-votes do not count. Counts below 10 are hidden for display only.
—Votes can only raise a score: up to 20% of it reflects upvotes, graded on a curve against the most-voted tool. The square-root curve has a floor of 100 effective votes. Votes decay with a 90-day half-life. Maker self-votes do not count. Counts below 10 are hidden for display only.
Rank history
Not ranked
First ranked 18 Nov 2026
First ranked 18 Nov 2026
First ranked 18 Nov 2026
Full page

Watch Crab Code vs Hermes Agent vs Codex CLI vs OpenCode on YouTube

More comparisons

Build your own