Project Brain

  1. › opening Project Brain
  2. › synthetic demo
  3. › loading workspace
Skip to content
Project Brainmemory for your AI agents

Workspace

DashboardProjectsMemoryQualityActivitySettingsSetup

Read-only example → install your own

Demo dataNumbers are examples until the API is connected.

Help

Settings/Access

Across all projects
Demo data · synthetic examples

Read-only preview. No repository or model is connected; the numbers and project knowledge are examples.

Install your own →

Loading this screen…

Access

Who can open this dashboard, and which agents and integrations can talk to Brain. Every person, agent and integration has its own with keys you can revoke. Each one only gets the it needs.

GeneralAccessAgent toolsReranker

Dashboard sign-in

The login and password for this web dashboard. Agents and integrations use their own keys below.

Demo dataNothing to protect yet

This dashboard shows example data. Sign-in appears here once it is connected to a Brain server.

Identities51 disabled
Active keys5keys that work right now
Expiring soon1within 14 days
Never used1active keys Brain hasn’t seen yet

New identity

Creates the identity and its first key. You’ll see the key once, right after.

+ New identityHide
Kind
Picks the starting permissions below; you can change them.
Permissions AI agent
core:writejobs:read
Choose permissions one by one

Write includes read in the same area.

Everyday
Code memorySearch code, build context packs, read and record decisions and rules.
Background jobsFollow jobs; write starts re-indexing, health checks and other jobs.
Agent task ledgerTasks the agent harness hands out and reports back on.
SetupSetup status; write changes the repository and model providers.
Insights
Code graphThe map of modules and calls; write rebuilds or rolls it back.
Architecture driftWhere code moves away from the agreed structure; write runs a scan.
CouplingParts of the code that keep changing together.
Change impactWhat a change touches — files, tests, call paths.
FreshnessHow current each repository's memory is; write rebuilds it.
PortfolioDependencies and cycles across several repositories.
Changes and experiments
Fix plansAssisted plans for fixing findings.
ExperimentsTrying fixes side by side before a person applies one.
Change labDisposable workspaces where patches are tried out.
Agent runsPhased agent execution sessions.
Shadow runsAgent runs replayed safely next to the real ones.
GoalsLong-running goals handed to agents.
ImprovementAgent tournaments and promotion of better setups.
RoutingWhich model or agent handles which request.read only
Oversight
Evidence ledgerThe tamper-evident record of what Brain did.read only
Audit logEvery change made through the API, for review or export.read only
Control queueDecisions waiting for a person.read only
OperationsNightly runs and incidents; write repairs vectors.
RepositoriesThe repository registry; write adds or disables repositories.
Admin
AccessIdentities and keys; write creates, revokes and disables them.
Team

These scopes do not create membership or a project role. Brain also checks the person’s live membership and project permissions; you can only mint scopes your current credential holds.

Connect the Brain API to run actions — this is demo data.

Key for “agent”

5 identities

Open one to see its keys, add a key or switch it off.

Loading list…
The master keyOlder integrations may still use PROJECT_BRAIN_API_KEY. It lives in the Brain server’s .env, can do everything and isn’t listed here. Check its recent requests under Activity before rotating it: existing clients stop working until their configuration is updated. Give each active integration its own key, verify it works, then change the master key on the server.
1–5 of 5

  1. New key for “admin”

    Key for “agent”

  2. New key for “claude-code”

    Key for “agent”

  3. New key for “codex”

    Key for “agent”

  4. New key for “ci-bridge”

    Key for “agent”

  5. New key for “old-laptop”

    Key for “agent”