Projects & Repositories
Associate approved repositories with projects. Manage repository access, Chat access and write-capable agent execution as separate entitlements.
Project
Repository
Read-only Chat
Write-capable CLI
BATONX / ADMIN CONSOLE
One control centre for AI development governance.
Centrally manage who can work with which projects, repositories, coding agents and models. Enforce policy, require approval for sensitive actions and produce verifiable evidence for every governed session.
Conceptual interface Healthy
Governed sessions
Claude CodeDeveloper A
Example project / example-app
CodexDeveloper B
Example project / example-app
Policy decisions
Allow01
Deny01
Approval required01
Pending approvals
Justification · expiry · recorded decisionEffective policyv3 · exampleModelApproved model
- Identity
- Project / repository
- Policy
- Decision
- Evidence
Configure
Approve resources. Define the scope.
Enforce
Apply policy to governed sessions.
Prove
Connect decisions to evidence.
Admin Console configures and governs the platform. BatonX CLI is the coding interface; BatonX Chat is the read-only project interface.
A conceptual view of the Admin Console. Capability scope and integration availability are confirmed for each pilot.
CENTRAL CONFIGURATION
Manage your organisation’s AI development controls from one place.
The Admin Console brings organisation, project and repository configuration into one governance model. Agents, runtimes, models and integrations are available only within explicitly authorised scopes.
Agents & Runtimes
Manage Claude Code, Codex and future agent adapters. Define approved runtime images, workspace modes and resource boundaries centrally.
Claude CodeCodex
Approved runtime boundary
Workspace modesResource boundaries
Providers & Models
Configure approved providers, endpoints, regions and models. Restrict model access according to organisation, project and repository policy.
Approved destinations
CloudPrivateInternal
PolicyPOLICY & APPROVALS
Define the rules once. Enforce them in every session.
Apply organisation policy and narrow it safely at project and repository scope. Define allow, deny and approval-required decisions for file operations, commands, network access, MCP tools and model usage.
- Organisation → Project → Repository policy hierarchy
- Risk-based command and operation controls
- Time-bound approvals with recorded justification
- Protection against weakening mandatory parent policies
- Versioned and audited policy changes
A model’s claim that an action is safe does not create authority. BatonX policies and technical controls make the final decision.
- OrganisationInherited mandatory controls
- ProjectScope may tighten permitted rules
- RepositoryScope may tighten permitted rules
Choose an illustrative outcome. This does not run an agent or change a policy.
Requested operation
Unapproved network routeEffective policy
BlockedAudit evidence
Blocked action and policy decision recordedIllustrative flow: inherited policy determines whether an operation proceeds, is blocked or awaits approval. Every decision connects to audit evidence.
Policy decisions
- Approved repository accessApproved operation proceeds → Access decision and scope recorded
- Unapproved network routeBlocked → Blocked action and policy decision recorded
- Sensitive operationHeld for human approval → Approval decision and justification recorded
LIVE OPERATIONS
Understand what is happening without collecting unnecessary content.
Monitor active sessions, agents, models, policy decisions, blocked actions, security events and platform health. BatonX provides operational visibility without centrally retaining source code, prompts or responses by default.
ActiveClaude Code
Example project
- Identity
- Developer A
- Repository
- example-app
- Model
- Approved model
- Effective policy
- v3 · example
PausedCodex / Developer B
Policy decisions
Network access blockedDeny → Decision recorded
Credential access blockedDeny → Decision recorded
Approval requiredJustification · expiry · recorded decision
System health
RuntimeHealthy
GatewayHealthy
Event deliveryDelivered
Tokens2.4k
Latency640 ms
Estimated cost$0.04
Illustrative session metadata and platform status. Source code, prompt and response content are not shown.
AUDIT & EVIDENCE
Preserve the evidence behind every critical decision.
Connect identity, project, repository, agent, model, policy, approval and security events through one traceable correlation chain.
BatonX / One traceable correlation chain
- Identity
- Project / repository
- Policy
- Decision
- Evidence
IdentityProjectRepositoryAgentRuntimeModelPolicyApproval required
Who performed the action, when and within which scope?
Which agent, runtime and model versions were used?
Which policy was applied?
Which actions were blocked or approved?
Can the integrity of the evidence be verified?
BatonX does not certify compliance. It produces verifiable technical evidence that supports your organisation’s audit and compliance processes.
USAGE & COST
Understand AI usage by project, provider and model.
Review requests, tokens, latency and estimated cost across organisations, projects, repositories, sessions, agents, providers and models. Apply soft and hard limits to control unexpected consumption.
Soft and hard consumption limits are part of the product direction; their scope is confirmed during pilot planning.
Example project
Approved provider / Approved model- Requests
- 12
- Tokens
- 18.6k
- Latency
- 640 ms
- Estimated cost
- $0.32
Soft limitThreshold notification
Hard limitConsumption boundary
These insights support security, capacity and cost management. BatonX does not score developer performance or rank employees.
EVENT OUTPUTS
Works with your operational systems.
BatonX emits standardised, versioned security and operational events. Route them into your existing webhook, SIEM, syslog, event-stream and observability infrastructure.
Email, SMS, Slack, Teams and incident-escalation workflows remain under customer control.
Output categories describe the integration architecture. Connector compatibility, including metrics and OpenTelemetry, requires integration-specific validation.
Explore integration scopeVersioned BatonX events
BatonX- HMAC-signed webhook
- SIEM and syslog
- Event streams
- Metrics and OpenTelemetry
- Customer-managed dashboards
ROLES & RESPONSIBILITY
The right level of control for every role.
Organisation Admins, Security Admins and Project Admins receive only the management capabilities required for their responsibilities. Sensitive operations can require separation of duties, additional approval and mandatory audit.
Organisation structure
Organisation Admin
Manages organisation structure, projects, user access and general platform configuration.
Security governance
Security Admin
Manages security policy, sensitive approvals, incidents, provider controls and protected audit evidence.
Authorised projects
Project Admin
Manages repository associations, user access and project-scoped configuration for authorised projects.
THE PRIVACY BOUNDARY
Security visibility, not employee surveillance.
BatonX is not an employee-surveillance or performance-management product. It does not rank developers using token consumption, prompt count, session duration or lines of code. Individual activity is available only where required for security investigation, access control, operations and audit.
Visibility for a defined purpose
- Security investigation
- Access control
- Operations
- Audit
Outside the product’s purpose
- Developer performance scores
- Employee rankings
- Productivity conclusions
BatonX / Admin Console
Bring AI coding-agent governance under your organisation’s control.
Let’s explore how BatonX can fit your identity, repository, model-provider and security environment.