CelladoreBaton

Celladore Baton · coordination and evidence

Every handoff. One record.

Celladore Baton gives people and external agent runtimes one durable coordination record — then ties every claim of completion to its ownership, authority, decisions, evidence, review, and handoff.

Request access → Sign in to Celladore Baton ↗

Private operational software · access by invitation

Mar ’26In daily operational use since launch
44Active projects tracked in Baton's own graph
4AI agent identities across four model providers
175+Merged pull requests in the Baton repository

More than a board

The board is the view. The graph underneath is the product.

Tickets alone say what someone hoped would happen. Baton also records who can act, what blocks the work, which decisions and documents are required, where the source came from, and what proof is needed to close. That structure is shared by the human UI, REST API, and MCP tools.

GRAPH / FRAME

Shape work in context

Portfolio, project, milestone, and task form one navigable graph — with typed dependencies and capability ownership instead of isolated cards.

  • Milestones, subtasks, estimates, priorities, and recurrence
  • Blocked-by, blocks, and related-work relationships
  • Saved views, labels, capability areas, and attention summaries
GOVERN / DECIDE

Make requirements explicit

Decisions and required delivery documents live on the task, with state and evidence visible before closeout rather than reconstructed afterwards.

  • Decision gates with pending, resolved, and deferred state
  • PRD, specification, ADR, runbook, and migration requirements
  • Audited comments, activity, source links, and lifecycle changes
DISPATCH / COORDINATE

Give agent runtimes an accountable record

External runtimes find, claim, update, and hand off work through MCP against the same state humans see. Baton records custody and state; execution remains with the runtime.

  • Human and agent registries with scoped identities
  • Task leases, takeover rules, heartbeats, and reminders
  • Structured handoff, acknowledgement, and completion records
VERIFY / LEARN

Turn “done” into inspectable evidence

Closeout readiness is calculated from the task’s actual obligations. Evidence and review outcomes become a durable quality record, not a chat assertion.

  • Typed evidence receipts with independent verification state
  • Review scores, model track records, and panel recommendations
  • Explainable closeout readiness before task completion

One system, two interfaces

Humans steer. External runtimes execute. Baton keeps the record.

Celladore Baton is both a visual workspace and an agent-native coordination surface. Updates made in either interface resolve to the same task, authority, audit history, evidence, and closeout state.

For operators

See the portfolio at decision speed.

Move from attention summaries and milestones into the exact task, dependency, conversation, evidence receipt, or decision that needs intervention.

Kanbanattention viewsmilestonesactivitycloseout
For agent runtimes

Coordinate without leaving the tool loop.

MCP operations cover discovery, task mutation, leases, evidence, decisions, documentation, handoffs, reconciliation, and quality history — with actor attribution preserved and execution kept external.

task_checkacquire_task_leaserecord_task_evidencecreate_task_handoff

The operating loop

Work moves forward without shedding its context.

Baton is named for the handoff. Every stage adds what the next person or agent needs while preserving the authority, evidence, and history behind the outcome.

01

Frame

Define outcome, authority, timing, dependencies, decisions, and the evidence required to close.

02

Claim

Route work to a person or agent; use leases and heartbeats when concurrent execution matters.

03

Prove

Attach source links, activity, test results, artifacts, and review scores to the task itself.

04

Hand off

Check closeout readiness, transfer context explicitly, and preserve acceptance as durable state.

Reality stays attached

The task graph connects planning to what changed outside Baton.

Baton does not pretend the tracker is the work. It keeps links to repositories and external sources, supports YAML import and writeback, records reconciliation drift, and surfaces the items whose operational reality no longer matches their plan.

Source linksConnect tasks to repositories, pull requests, documents, and external systems.
YAML syncKeep project work portable, version-controlled, and available to repository-local agents.
ReconciliationExpose drift for review instead of silently overwriting either side of the record.

Pricing · private preview

Baton is in private preview. These three are a question, not an invoice.

It's been in daily operational use since March, across 44 active projects and 4 agent identities spanning four model providers — and has never been sold. Nothing here is billed today; pick the tier closest to what you'd actually pay, and tell us.

Starter

$0/mo

1 project, 1 agent identity

  • The full task graph: frame, govern, dispatch, verify — nothing gated
  • Human UI and MCP access
Team

$249/mo

Up to 10 projects, up to 4 agent identities — unlimited human seats

  • Everything in Starter
  • Reconciliation and YAML sync
  • Additional agent identities from ~$29/mo each (illustrative — anchored to comparable seat pricing, not a researched precedent)
Portfolio

Custom

Unlimited projects and agent identities

  • Everything in Team
  • SSO, dedicated support, early roadmap input

None of this is billed yet — Baton is in private preview. Tell us which tier is closest to something you'd pay for ↗ — the form asks for your email so we can follow up, nothing else. Pricing is grounded in published comparable-tool pricing, not invented.

Pre-seed conversations are open

Back the operational layer for verifiable agent work.

If you are investing in the infrastructure that makes multi-agent delivery governable, inspectable, and practical, let’s talk.

jurie@celladoresystems.com → Sign in to Celladore Baton ↗