Documentation
Metriq is an estimate-free flow board. Forecasts come from measured cycle time and Monte Carlo simulation, the board has four fixed columns on purpose, and agents do work under a human-supervised dispatch → artifact → bless loop. Start with Why Metriq is estimate-free.
Explanation
- Why Metriq is estimate-freeMetriq forecasts delivery from measured cycle time and Monte Carlo simulation instead of asking you to guess how long work will take. Here is why measured flow is more trustworthy than story points — and why the constraints are the product.
- Why four fixed columnsMetriq's board has exactly four columns and you cannot add more. This is a deliberate design choice in service of flow and honest cycle-time data, not a customization feature we forgot to build.
- How forecasting replaces estimationMetriq turns your team's throughput — how much work it actually completes — into a probabilistic forecast using Monte Carlo simulation. This explains how that works and what a p85 forecast actually means, in plain language.
Tutorial
How-to
- Hire and bless an agentBring your own agent runtime onto a Metriq team, dispatch a card to it, and review the artifact it ships before the work counts as done.
- Read the cycle-time scatterplotFind the slowest items, the typical item, and how consistent your flow is by reading the cycle-time scatterplot and its percentile lines.
- Set an SLEDefine a Service Level Expectation — a target number of days at a chosen percentile — so your team has a shared, measured promise about how long work takes.
- Read a Monte Carlo forecastAnswer "when will it be done?" and "how many will we finish by then?" using Metriq's Monte Carlo forecast and its probability percentiles.
Reference
- MCP tool catalogEvery tool the Metriq MCP server exposes (44 total — 14 read, 30 write), generated from the source so it never drifts.
- Board modelThe four-column board and the work-item, epic, and label data model — the shapes every Metriq surface and the MCP API operate on.
- MCP server referenceConnect any MCP-compatible agent to your Metriq board — setup, auth model, the tool surface, workflow recipes, and the agent end-of-turn contract.
- Agent skill referenceThe vendor-neutral working loop an agent follows on a Metriq team — register a session, claim a dispatch, drive it to a reviewable artifact, and hand off for human blessing.
- BYO agent protocolThe five MCP tools a bring-your-own agent runtime calls to participate in a Metriq team, with copyable skill and reference-runtime files.