Glossary
The terms Leanmote works with, each defined in a few lines: what it means and how Leanmote uses it.
Agent harness
An agent harness is everything around an AI coding agent that decides how well its work ships: the context files it reads, such as AGENTS.md, its prompts and rules, its tools, and the review, integration and deploy steps its output passes through.
AI Return
AI Return is how much of an investment in AI tools actually turns into shipped software.
Baseline
A baseline is the normal range of a team's metrics, calculated from its own history, against which every later change is measured.
Cost of rework
The cost of rework is what redoing work costs in money: the hours people spend on it, at their real hourly cost, plus the AI spend consumed by the first attempt and by the second.
Cost per outcome
Cost per outcome is what a team spends, in people and in AI tools, for each unit of work that actually ships.
Cost per PR
Cost per PR is the engineering cost of a period divided by the pull requests merged in that period.
Cycle time
Cycle time is how long an item takes from the moment active work starts to the moment it is done.
DORA metrics
DORA metrics are the four standard measures of how well an engineering team delivers software: deployment frequency, lead time for changes, change failure rate and time to restore service.
Flow metrics
Flow metrics measure how efficiently work moves through a team's delivery system before deployment.
Leak
A leak is work or spend that goes nowhere: rework, queues, stalled items, and AI seats and tokens without a return.
Leak pricing
Leak pricing means putting a dollar amount on each operational problem, so it can be ranked and funded like any other cost.
Metrics Context Layer
A Metrics Context Layer is one per-team view of an operation's metrics, built from the tools a company already uses and calibrated on each team's own history, that AI assistants can query directly.
Resolution agent
A resolution agent is an AI agent that owns one specific problem pattern in a team's workflow, such as code review, rework, queues or deploy risk, and executes the action that pattern calls for.
Rework
Rework is work that has to be done again after it was considered done or ready: a ticket that bounces back, a pull request that returns from review, a change that breaks and has to be fixed.
Waiting time
Waiting time is the time work items spend in non-active states: waiting for a review, a decision, another team or a deploy.
WIP (work in progress)
WIP, or work in progress, is the number of items that are started but not finished at a given moment.