Two-record completion checks
Compare what an agent says it did with what the host actually observed, turning disagreements into findings.
Yoetz
Freemium
Yoetz is a local-first, open-source work ledger that checks completion claims from AI coding agents. It compares the agent’s declared work with host-side records and issues a receipt showing what was verified, the coverage reached, and any open findings.
What is Yoetz?
Yoetz is a local-first work ledger for AI coding agents. An agent publishes its plan, work, claims, and evidence; Yoetz compares that record with host-side observations such as sessions, tool calls, and edits. When the records disagree, the disagreement becomes a finding rather than being hidden behind a simple “done” status.
The workflow produces a receipt for each completion claim. A receipt states what was verified, the coverage reached, and what remains open. The example workflow shows Yoetz identifying a test run that predates a later edit and an API response shape that was not verified, then preserving unresolved findings when the final receipt is issued.
Yoetz runs as one local service and uses MCP for agent communication and hooks for host reporting. The site describes the host record as structural facts rather than file contents. The product is officially supported for Codex, Claude Code, and Cursor, and the site presents it as open source under the Apache-2.0 license.
The homepage presents setup through an agent prompt that installs Yoetz and shows changes before proceeding, as well as manual installation options using PyPI or npm. The documented platform status is stable on macOS, while Linux and Windows through WSL 2 are described as experimental. The product currently has no account or network requirement according to the site, and is identified as alpha software at version 0.2.5.
What can Yoetz do?
Coverage-bounded receipts
Issue a receipt that identifies verified obligations, the checking coverage, and unresolved findings instead of treating every completion claim as equally confirmed.
MCP and hook-based reporting
Let the agent publish its plan, work, claims, and evidence over MCP while Codex, Claude Code, and Cursor report sessions, tool calls, and edits through hooks.
Append-only local ledger
Keep both records in one local service so local rules can compare them without sending the ledger to a remote service.
Agent-guided setup
Use a copied prompt to install Yoetz with an agent and review the changes first, or choose the manual PyPI and npm installation paths.
Use Cases
“Validate an AI-assisted refactor”
Ask an agent to change code and preserve an interface, then use the check to expose stale test evidence or an obligation that was never verified before accepting the work.
“Review completion claims in an individual repository”
Use the local ledger to maintain a record of plans, edits, tool activity, and evidence while an agent works without requiring an account or network connection.
“Inspect unresolved findings”
Use the receipt as a handoff or review artifact when a check finds problems, including cases where the final claim still carries open findings.
“Coordinate early multi-agent work”
The site describes an alpha workflow for delegated child tasks, receipt rollups, and consent-based notices when two live tasks touch the same path; a clean child receipt does not prove integration.
Frequently Asked Questions
What does Yoetz check?
It compares the agent’s published record of its plan, work, claims, and evidence with host-side records such as sessions, tool calls, and edits. Disagreements are reported as findings.
What does a Yoetz receipt contain?
The receipt states what was verified, the coverage reached, and what remains open. The homepage example includes verified obligations and an unresolved finding for evidence that was stale or incomplete.
Which coding agents are officially supported?
The site lists Codex, Claude Code, and Cursor as officially supported integrations. Yoetz communicates with the agent over MCP and receives host-side reports through hooks.
Which platforms does Yoetz support?
The site describes macOS as stable. Linux and Windows through WSL 2 are described as experimental.
Does Yoetz require an account or network connection?
The homepage says Yoetz works today with no account and no network. It describes the product as local-first, with one local service holding the records and performing the checks.
Quick Facts
- Product type
- AI coding-agent completion checker and local work ledger
- Primary users
- Developers using AI coding agents
- Official integrations
- Codex, Claude Code, and Cursor
- Platform status
- macOS stable; Linux and Windows via WSL 2 experimental
- License
- Apache-2.0 open source
- Current status
- Alpha, version 0.2.5
Yoetz Alternatives
Catenary
thecatenary.app
Catenary is a local-first desktop IDE for running and coordinating multiple AI coding agents on an infinite canvas. It helps developers keep terminals, editors, browsers, worktrees, and agent context visible in one workspace.
Bentomux
bentomux.takora.dev
An open-source desktop app for running coding-agent CLIs side by side in persistent workspaces.
SuperQode
superqode.dev
SuperQode is an open-source harness layer for coding agents that lets users discover, build, run, evaluate, and optimize agent harnesses from a single terminal. It supports interoperability across A2A, ACP, MCP, and UHP, so teams can keep using their existing coding agents and models.
Weave Router
weaveos.com
Weave Router is an intelligent model router for coding agents. It evaluates coding tasks and routes requests across open and closed models, helping developers keep existing agent workflows while managing speed, quality, and model spend.
Experiential Labs
www.experientiallabs.ai
Experiential Labs is an open-source AI gateway that gives teams one endpoint and API key for hosted models, customer-owned keys, and local or custom models. It adds routing, budgets, access control, attribution, and optional intelligence features for agent traffic.
ResumeContext
resumecontext.com
ResumeContext is a shared memory service for coding-agent teams. It records agent sessions locally, syncs them to a project archive, and exposes that archive to agents through MCP so they can search prior work and continue across agents, accounts, or machines.