ContextCode v0.6.57 · macOS · Linux · Windows

Every coding agent starts from zero.This one starts from everything you’ve already decided.

ContextCode is the terminal coding agent wired straight into ContextStream. Your team’s decisions, corrections and lessons ground the plan before the first edit — and whatever this turn learns lands back in the same memory when it ends.

curl -fsSL https://contextstream.io/cs/install.sh | sh

01 The engine

The keyboard, and the memory.

Every turn opens by asking the workspace what is already known and closes by giving back what it learned. That round trip is the whole difference.

1
One sign-incsc login approves the device in your browser. No API keys to paste, no config file to write.
2
One scopeWorkspace and project resolve from the repo itself, so the memory you get in this checkout is the memory that belongs to it.
3
One balanceCredits are shared across terminal, dashboard and desktop — and your own provider keys run at zero credits.

The same workspace every other assistant reads through the MCP server. ContextCode is the first-party path to it, not the only one.

02 The proof

Memory was the only variable.

8 tasks against one repository, 3 trials per setup, every session fresh. Same Claude Sonnet 4.6, same prompts — the only change was whether project memory was connected.

Agent task success

July 2026 · result of record

48 scored runs
No ContextStreammodel + repo only
14/2458%
ContextStream memoryindex + project memory
23/2496%
+9more tasks completedSame Claude Sonnet 4.6. Same codebase. Fresh sessions.
Knowledge not in code
0/9without memory9/9with ContextStream
48 scored runs · 8 tasks · 3 trials per setupEvery result, the methodology, and our LongMemEval run →
03 What it does

A teammate, not a text box.

Grounding is what makes the work good. These are the things it does with it.

A team, safely parallel

One goal, fanned out across coders, a reviewer and an integrator.

Isolation
a git worktree per agent
Landing
read the diff, Enter merges
/orchestrate

Verified, not asserted

A real browser clicks through the feature the way a user would.

Proof
PASS or FAIL, with a screenshot
Loop
compiler errors feed the edit
/verify

Judged by your own rules

Plans and diffs are both measured against the decisions this repo is already bound by.

Binding
recorded decisions and lessons
Verdict
severity, file:line, the fix
/plan · /review

The right model per step

Reasoning for the plan, a fast coder for the edit — routed for you.

Override
one keystroke, mid-session
Cost
your own keys run at zero
Ctrl+M

It runs your actual stack

Rust, Go, TypeScript, Python, Ruby, C# and PHP, with your real toolchain.

Honest
it probes a URL before citing it
Search
meaning-aware, not grep
csc run

Headless, in CI, over MCP

The binary you type into is the one your pipeline calls. No TTY, no prompts.

Envelope
one JSON object per run
MCP
a client, and a server
csc run --json

Convinced by the engine? Install it on one real repo and ask it something your code cannot answer.

Get ContextCode
04 The staff

Not an assistant. A staff.

Seven specialists, each with its own model, budget and permissions — bound in a JSON manifest rather than asked for politely in a prompt. Drop your own in .contextcode/agents/ and it resolves ahead of the built-ins.

  • leadRuns the teamThe only specialist that can spawn others, and the one /orchestrate puts in charge.
  • plannerDecides what to doFive tasks or fewer, no essays — grounded in the decisions, lessons and runbooks already recorded.
  • coderImplements the changeRuns your real build and feeds its own failures back into its next edit.
  • reviewerJudges the diffRead-only by construction — it cannot write a file, whatever a prompt asks it to do.
  • explorerAnswers “how does this work?”Investigates code and memory at once, and surfaces the runbook that already answers the question.
  • designerDesigns the interfaceGenerates real images, builds the screen, then opens a browser and looks at what it made.
  • prototyperMakes it real, fastThe widest turn budget of any specialist, tuned for something you can click rather than read.
05 Head to head

Where they diverge.

Claude Code and Codex CLI are both capable tools. The difference is what happens around the model — and it works here on day one.

built in partial / manual setup not available
Capability comparison between ContextCode, Claude Code and Codex CLI.
CapabilityContextCodeClaude CodeCodex CLI
Context & memory
Persistent memory of decisions and lessonsBuilt in. Binding on every turnPartial, or manual setup. Local filesPartial, or manual setup. AGENTS.md only
Memory shared across machines and teammatesBuilt in. One workspace brainNot available. Per-developerNot available. Per-developer
Recall past sessions — “what did we decide?”Built in. Indexed and searchablePartial, or manual setup. Resume onlyPartial, or manual setup. Resume only
Agents & isolation
Multi-agent orchestration with budgetsBuilt in. A dependency graphPartial, or manual setup. Subagents, per promptPartial, or manual setup. Cloud tasks
Git worktree isolationBuilt in. Default-onPartial, or manual setup. Optional, agents onlyNot available. Manual git
Review-first, one-key merge of agent workBuilt in. Diff card → EnterNot available. Merge it yourselfNot available. Merge it yourself
Verification
Browser verification with screenshotsBuilt in. Auto-provisionedPartial, or manual setup. Via MCP you set upPartial, or manual setup. Via MCP you set up
Self-correcting edits — compiler feedbackBuilt in. After every edit batchPartial, or manual setup. Via hooks you writePartial, or manual setup. Manual test runs
Models & pricing
Multiple frontier providers in one toolBuilt in. Routed per stepNot available. Anthropic modelsNot available. OpenAI models
Honest usage pricingBuilt in. Pay for what you usePartial, or manual setup. Weekly plan limitsPartial, or manual setup. Weekly plan limits

All three ship MCP client and server modes, AGENTS.md, custom slash commands, skills, hooks, resume and continue, image paste, headless runs — the table is only where they diverge, and it reflects publicly documented capability as of July 2026. Both are excellent tools; if a cell looks out of date, tell us and we’ll fix it.

06 Every surface

Same agent, three surfaces.

The dashboard and the desktop app are not reimplementations — they run the released binary. One memory, one credit balance.

csc · ~/dev/api
A keyboard.

Your terminal

macOS · Linux · Windows

The full agent — orchestration, worktrees, verification, review — as one static binary.

Read the docs
contextstream.io · ContextAgent
A browser.

Your dashboard

Any browser, including your phone

The same released binary, driven from the web. Start a task at the keyboard and it is waiting here.

See the workspace
ContextStream.app · ContextAgent
A window.

Your desktop

macOS · Windows

It bundles the binary too. Ask what the project decided without opening a terminal at all.

Download
07 Install

Sixty seconds to grounded.

One static binary. No Node, no Python, no plugin. It works alongside VS Code, JetBrains, Neovim or whatever else you already have open.

Run thisLatest · v0.6.57
macOS/Linuxcurl -fsSL https://contextstream.io/cs/install.sh | sh
Windowsirm https://contextstream.io/cs/install.ps1 | iex
1
InstallOne command, checksums verified. csc update keeps it current from signed releases.
2
csc loginYour browser opens; approve the device. Done.
3
csc in any repoYour first prompt is already grounded. Try /plan on something real.
Safe by default — agents work in isolated git worktrees. Nothing lands on your branch until you press Enter, and /undo takes back any turn.
Included with Pro, Elite and Team — from $19/month, on the same credit balance as the dashboard and desktop app.
Direct binaries and checksums

Explain it once. Let it compound.

The first plan you approve is already part of your team’s memory — and part of what the next agent starts from.