brain0 is live · free plan; paid price not published · last verified 25 Oct 2026 · viability healthy
brain0
brain0 is an open-source decision graph that links every git commit to the coding-agent prompts and reads behind it.
Drift reconciliation and the DLP read audit answer questions a git log genuinely cannot, and packaging them behind one `npx` command with no agent cooperation is the interesting part. Add the MCP server and your own repository history becomes retrieval memory for the next agent session. Skip it if humans still write most of your commits — there is no why layer to reconstruct, and you'll be staring at a graph of empty intents.
Verdict generatedVerified 5d ago · liveness 79/100 · cite: rightaichoice.com/tools/brain0
- Engineering teams whose diffs mostly come from Codex or Claude Code and need a why layer on top of git
- Security and platform teams auditing which paths and secret kinds agents read before a push
- Developers doing root-cause work on a bug an AI agent introduced weeks ago
- Open-source maintainers who want provenance for AI-assisted contributions
- Teams that don't use AI coding agents — the intent side of the graph stays empty
- Anyone expecting real-time interception of a running agent rather than after-the-commit review
- Non-technical stakeholders who want a business-facing dashboard instead of a graph explorer
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip brain0 if you want an agent intercepted mid-session rather than reviewed after the commit, or if your team writes code by hand — there is no prompt layer to reconstruct and the graph will look empty.
Semantic search and GUI smart chat need a first run with Ollama (or an Anthropic/OpenAI key) to populate the summary and embedding cache — until then you are on deterministic summaries only.
brain0 is Apache-2.0 and free at $0 — there is no per-seat meter and no usage tier to graduate into, so cost does not scale with headcount the way it does with hosted agent-observability suites that price per developer or per GB of ingested transcript. The trade is operational rather than financial: you supply Node.js 20+, the machine that runs the CLI, and optionally a local Ollama install.
In short
brain0 — brain0 is an open-source decision graph that links every git commit to the coding-agent prompts and reads behind it. Best for Engineering teams whose diffs mostly come from Codex or Claude Code and need a why layer on top of git, Security and platform teams auditing which paths and secret kinds agents read before a push, Developers doing root-cause work on a bug an AI agent introduced weeks ago. Free to start; it also has paid plans whose price we could not confirm — see the pricing table.
What is brain0?
brain0 — brain0 is an open-source decision graph that links every git commit to the coding-agent prompts and reads behind it. Best for Engineering teams whose diffs mostly come from Codex or Claude Code and need a why layer on top of git, Security and platform teams auditing which paths and secret kinds agents read before a push, Developers doing root-cause work on a bug an AI agent introduced weeks ago. Free to start; it also has paid plans whose price we could not confirm — see the pricing table.
brain0 is an open-source tool for repositories whose diffs are increasingly written by coding agents. It passively builds a decision graph of your codebase: git holds the facts, the transcripts Codex (~/.codex) and Claude Code (~/.claude/projects) already write to disk hold the intents, and brain0 links each commit back to the prompt behind it, down to a single function. No hooks, no agent cooperation, no code changes are required — it just reads what's already on the machine. Setup is `npx brain0 up` from any repo, with Node.js 20 or newer as the only hard requirement (Linux needs glibc 2.28+; musl/Alpine is listed as unsupported for now). The command infers your repo id from the git remote, indexes history and sessions, and opens a GUI graph explorer at localhost:8787 where you drill from repo to module, file and symbol. The features that matter are specific. Drift detection reconciles what an agent declared it changed against what git actually recorded and scores the gap. A DLP audit logs which paths each session touched and what kind of secret they held — never the values. A two-dimensional risk score combines blast radius and churn with later evidence such as reverts and follow-up fixes, rendered green to red. Daily use runs through `brain0 today` for triage, `brain0 report --md` for a shareable accountability doc, and `brain0 query "why did the parser break"` for root cause. Offline is the default posture: with nothing else installed you get deterministic summaries, local feature-hash embeddings and zero egress. Adding Ollama (qwen3:4b, qwen3-embedding:0.6b) unlocks model-written summaries, semantic search and GUI smart chat at no cost; hosted chat is opt-in via an API key. An MCP server hands the graph to coding agents as memory, and signed provenance attestations cover compliance review. Compared with observability suites aimed at hosted agent platforms, brain0 is local-first and Apache-2.0 licensed, and the tradeoff is a narrower surface: it reads what
Key features
- Passive decision graph linking each commit to the agent prompt behind it, down to the function
- Drift detection scoring the gap between agent-declared changes and git-recorded diffs
- DLP audit logging which paths each session read — secret kinds only, never values
- Captures explicit reads, Grep results, shell reads like cat .env, and subagent transcripts
- Two-dimensional risk score fusing blast radius and churn with reverts and follow-up fixes
- One-command setup with npx brain0 up — no hooks, no agent cooperation, no code changes
- Auto-discovers Codex (~/.codex) and Claude Code (~/.claude/projects) sessions
- GUI graph explorer at localhost:8787 from repo to module, file and symbol
- CLI workflows: brain0 today, brain0 report (--md), brain0 query, brain0 watch
- Offline-first by default: deterministic summaries, feature-hash embeddings, zero egress
- Local LLM support via Ollama (qwen3:4b summaries, qwen3-embedding:0.6b embeddings)
- Opt-in hosted LLM for GUI smart chat via ANTHROPIC_API_KEY or OPENAI_API_KEY
- MCP server exposing the decision graph to coding agents as memory
- Signed provenance attestations for compliance and legal review
- Filesystem checkpoint mode via brain0 watch
Behind the verdict
Where this earns its place is root-cause work on code an agent wrote weeks ago. A reviewer asks why a parser broke; git shows you the line, brain0 shows you the session, the prompt, and whether the agent's own account of the change matched the diff. That reconciliation is the feature I'd install it for, and I haven't seen drift scoring packaged this way in a general code-review tool. The security angle is the other reason teams will pull it in. The DLP audit tells you which paths a session read and what secret kinds appeared — .env, private keys — without storing values. If your worry is 'did an agent read credentials before pushing,' this is the cheapest way to find out on existing history. Treat it as a lower bound on what entered context, not a complete record. Pick brain0 when your diffs are mostly agent-written and you want a local, passive record you can query. Pass if humans still author most commits — the intent side stays empty and you're left with a graph of code only. Also pass if you need real-time interception of a running agent; this is after-the-commit review. In practice the first `npx brain0 up` is quick, but the payoff grows with history: a repo with months of Claude Code sessions gives you far more than a fresh one. On modest hardware, run the lighter summarizer for the initial cache-populating pass; summaries are content-cached, so later rebuilds cost nothing. The closest alternative is treating your agent platform's own logs as the source of truth. That works until you switch agents, or until the logs age out of the vendor's retention. brain0 keeps the record in your repo's orbit and reads both Codex and Claude Code. The caveat is maturity — 323 stars, 13 commits, Apache-2.0, and no enterprise support contract behind it.
Models under the hood
as of 2026-10-22
Limitations
- brain0 is a passive, post-hoc audit tool: it links commits to the agent prompts behind them rather than monitoring agents in real time, so it will not stop a session mid-flight.
- Session auto-discovery targets Codex (~/.codex) and Claude Code (~/.claude/projects), so other agent harnesses won't produce the prompt-to-commit "why" layer.
- Setup is one command (npx brain0 up) with no hooks or agent cooperation required, and it is offline by default with optional Anthropic or OpenAI keys only for hosted GUI smart chat.
- The DLP audit logs paths and secret kinds only, never values, and should be read as a lower bound on model context.
as of 2026-10-15
Researching brain0? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Is brain0 still active?
brain0 is live · free plan; paid price not published · last verified 25 Oct 2026 · viability healthy.
Viability score
How well maintained and how widely used is brain0? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: October 2026
How we score →What do users say about brain0?
What people actually say about brain0 — is it worth it?
Our community sentiment research written Jul 16, 2026 is more than 90 days old, so it is being re-reviewed against current data and is not shown.
How we verified this
Every listed fact about brain0— pricing, features, links, status — is re-read from the vendor's own pages by our verification pipeline and each change is logged with a date; the last pass ran on 25 Oct 2026. Pages mix model-assisted synthesis with human review; see how we verify.
Verification history
We have re-verified brain0 10 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — changed features, pricing tiers, summary, description, our verdict, who should skip it
- — changed pricing page
- — changed our verdict, who should skip it, who it suits, features, integrations, summary, pricing tiers, description
- — changed pricing page, models, who should skip it, our verdict, use cases, setup time, pricing fit, hidden costs, switching in, description, switching out
- — changed features, who should skip it, summary, description, our verdict, pricing tiers, who it suits
- — changed limitations, pricing page, models
Showing the 6 most recent of 10 verification passes.
Free to cite with attribution — this page re-verifies continuously.
How much does brain0 cost in 2026?
brain0: Free to start; it also has paid plans whose price we could not confirm — see the pricing table. Pricing last verified 25 Oct 2026.
Plans compared
Every published brain0 tier, its price as the vendor publishes it, who it fits, and what it adds over the tier below. The 12-month cost projection follows.
| Plan | Price | Billing | Who it's for |
|---|---|---|---|
| Open source (Apache-2.0) | Free | — | — |
| Local models (Ollama, self-hosted) | Free | — | — |
| Hosted LLM (opt-in, bring your own key) | Your provider's usage rates | — | — |
12-month cost
Project the real annual outlay, including the implied monthly cost when only an annual tier is published.
Vendor list price only. Add-on usage, seat overages, and contract minimums are surfaced under Hidden costs & gotchas.
Where the pricing makes sense
The company stage and team size where brain0's pricing actually pencils out — and where peers do it cheaper.
brain0 is Apache-2.0 and free at $0 — there is no per-seat meter and no usage tier to graduate into, so cost does not scale with headcount the way it does with hosted agent-observability suites that price per developer or per GB of ingested transcript. The trade is operational rather than financial: you supply Node.js 20+, the machine that runs the CLI, and optionally a local Ollama install.
Has brain0's pricing changed?
No plan price for brain0 has moved since we began logging in Aug 2026; the pricing page is re-read on every verification pass.
Who should pick brain0?
Pick brain0 if you are Engineering teams whose diffs mostly come from Codex or Claude Code and need a why layer on top of git, Security and platform teams auditing which paths and secret kinds agents read before a push, Developers doing root-cause work on a bug an AI agent introduced weeks ago, Open-source maintainers who want provenance for AI-assisted contributions.
Look elsewhere if you are Teams that don't use AI coding agents — the intent side of the graph stays empty, Anyone expecting real-time interception of a running agent rather than after-the-commit review, Non-technical stakeholders who want a business-facing dashboard instead of a graph explorer, Organizations treating the DLP read log as proof of model context rather than a lower bound.
Skip it if Skip brain0 if you want an agent intercepted mid-session rather than reviewed after the commit, or if your team writes code by hand — there is no prompt layer to reconstruct and the graph will look empty.
GeneratedReal-world workflow fit
Concrete scenarios for the personas brain0 actually fits — and what changes day-one when you adopt it.
Run `brain0 today` first thing with coffee, before opening any diffs. The list is ordered riskiest-first by the two-dimensional score, so the twelve-file commit the agent undersold floats to the top instead of hiding in the middle of a PR queue.
Outcome: You review the four commits that actually carry blast radius, kick the rest back to the agent, and skip reading a night's worth of low-risk churn line by line.
Run the DLP audit over the session range in question. brain0 lists which paths each session read — explicit Read calls, Grep hits, shell reads like `cat .env` — recording secret kinds but never the values themselves, then hand the path list to the incident review.
Outcome: You produce a defensible answer about what the session touched without making a second copy of the secrets, and you can say plainly that the log is a lower bound on context.
Run `brain0 query "why did the parser break"` instead of bisecting a thousand-line diff. The graph walks from the failing symbol back to the commit and then to the prompt and context that produced it.
Outcome: Root cause arrives in minutes with the originating prompt attached, and the post-incident write-up writes itself from `brain0 report --md`.
Use cases
- Trace an AI-generated commit back to the exact prompt and context that produced it
- Catch an agent that touched twelve files while declaring it changed two
- Audit which paths a Codex or Claude Code session read before a possible secret exposure
- Triage a morning of agent output riskiest-first with brain0 today
- Ask "why did the parser break" over the graph instead of reading a thousand-line diff
- Produce a shareable accountability report (brain0 report --md) for a post-incident review
- Give a coding agent the repo's decision graph as memory via the MCP server
- Attach signed provenance attestations to AI-assisted commits for compliance review
Setup time & first value
How long it actually takes to get something useful out of brain0 — broken out by persona, not the marketing-page minute.
From any repo, `npx brain0 up` is the whole install — no hooks, no agent cooperation, no code changes — and it opens the GUI at localhost:8787 right after the first index. Plan on minutes for a small repo and noticeably longer for a large monorepo's full git history. Optional Ollama models (qwen3:4b, qwen3-embedding:0.6b) add a one-time download plus a cache-populating run before semantic search
Switching to or from brain0
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From plain git log archaeology: run `npx brain0 up` in the repo, let it index existing history, and read the graph instead of grepping commit messages for the change that broke things
- →From a hosted agent-observability suite: stop shipping agent transcripts off-machine and point brain0 at the same Codex (~/.codex) or Claude Code (~/.claude/projects) transcripts that already sit on disk
- →From manual PR review of agent output: keep your existing review flow but sort it with `brain0 today` and let drift detection flag commits where the declared change and the recorded diff disagree
- →From a compliance spreadsheet of AI-assisted commits: use signed provenance attestations from brain0 instead of hand-maintained records
- ↗To a runtime agent gateway: keep brain0 as the historical record while the gateway handles in-session interception, since brain0 is post-hoc by design
- ↗To a hosted agent-observability platform: export accountability reports with `brain0 report --md` before moving transcript ingest off-machine
- ↗To manual git review only: stop running `brain0 watch` and delete ~/.cache/brain0 to drop the local index and checkpoints
Integrations
Resources & guides
- Resourcegithub.com
Brain0 · brain0
Helpful link from github.com
- Resourcegithub.com
README · brain0
Helpful link from github.com
- Resourcegithub.com
CONTRIBUTING · brain0
Helpful link from github.com
- Resourcegithub.com
SECURITY · brain0
Helpful link from github.com
- Resourcegithub.com
CODE OF CONDUCT · brain0
Helpful link from github.com
Tutorials & Learning
YouTube returned 6 videos for “brain0”, and we withheld 6: 6 could not be judged, because “brain0” is a single word that other videos use for other things. We are showing none, because we could not prove any of them are about brain0.
Official links
What are the best brain0 alternatives?
The closest alternatives to brain0 we list are Skylos, guard-skills and Qodo. Ranked by direct product-type match, not category overlap.
| Alternative | Pricing | Starts at | Last verified |
|---|---|---|---|
| Skylos | Freemium | Pricing not published | — |
| guard-skills | Free | Free | — |
| Qodo | Freemium | Pricing not published | — |
Alternatives to brain0
View all
Skylos
Skylos is an open-source PR check that catches AI agents deleting tests, skipping tests, or hard-coding answers before they merge.
guard-skills
Five free, open-source quality-gate skills that catch hallucinated APIs, assertion-free tests, and stale docs in AI-generated code before your agent commits
Tools that pair well with brain0
Common stack mates teams adopt alongside brain0, with the specific reason each pairing earns its keep.
Skylos
Skylos is an open-source PR check that catches AI agents deleting tests, skipping tests, or hard-coding answers before they merge.
guard-skills
Five free, open-source quality-gate skills that catch hallucinated APIs, assertion-free tests, and stale docs in AI-generated code before your agent commits
Qodo
Agentic AI code review that enforces your team's standards on every pull request and inside your coding agents.
brain0 vs its closest rival
brain0's closest head-to-head is with Cognition Ai — For large enterprises needing an autonomous engineer that ships production code, Cognition AI's Devin (with its new Security Swarm and Productivity Guarantee) is the clear choice. If your need is auditing and tracing AI-generated code back to prompts, brain0 is free, offline-first, and invaluable for compliance. They solve completely different problems, so choose based on whether you want to automate coding or audit it. Verdict generated
Head-to-head comparisons
Brain0 vs Cognition Ai
For large enterprises needing an autonomous engineer that ships production code, Cognition AI's Devin (with its new Security Swarm and Productivity Guarantee) is the clear choice. If your need is auditing and tracing AI-generated code back to prompts, brain0 is free, offline-first, and invaluable for compliance. They solve completely different problems, so choose based on whether you want to automate coding or audit it.
GeneratedBrain0 vs Bito
If you manage large multi-repo codebases and need AI coding agents to understand system-wide context for accurate code generation and architectural planning, Bito is the clear choice. If your priority is auditability — tracing every commit back to the AI prompts that produced it — and you prefer an offline, open-source tool, pick brain0. For most teams, Bito’s live knowledge graph and AI Architect capabilities offer deeper productivity gains, whereas brain0 is essential for compliance and transparency.
GeneratedFrequently asked questions
Best-of guides
Used brain0? Help shape our editorial sentiment research.