Expand description
Validated commands and consistent execution projections for humans and agents.
Structs§
- Affected
Work - Work whose execution context contains a replaced decision or its replacement.
- Aging
Work - One task in progress and how long it has been held.
- Applicability
Change - Work whose membership in the active graph a proposal changes.
- Applied
Calibration - The factors a calibrated
statusapplied to a copy of the plan; the stored plan and its estimates are unchanged. - Applied
Factor - The estimate factor for one executor kind.
- Applied
Review Delay - The review wait added for tasks still awaiting verification.
- Basis
Report - Derived validity of recorded provisional bases, read from the snapshot’s attempt records.
- Calibration
Report - Calibration of estimates and waits, and flow metrics, at one clock reading.
- Calibration
Rules - Thresholds and units of a calibration report.
- Change
Preview - Validated, read-only review of a proposed plan against one revision.
- Claim
Outcomes - Counts of claim episodes by how they ended.
- Durations
- Distribution of elapsed durations.
- Entity
Change - One entity-level semantic difference, including explicit additions and removals.
- Estimate
Calibration - Actual against estimated execution time.
- Estimate
Sample - One verified task’s actual execution time against its estimate.
- Exclusion
- Items left out of one measurement for one reason.
- Execution
Context - Resolved domain records needed by an agent to execute a work contract without guessing references.
- External
Link Request - Link request recording the resolved reference identifier so the operation log is explicit.
- Flow
Report - Flow metrics at one clock reading.
- Forecast
Spread - Two measures of the finish’s spread, which differ wherever near-critical paths matter.
- Gate
Report - Shared eligibility of one lifecycle transition for queries, commands and terminal projections.
- History
Coverage - Extent of the operation log the report read.
- Holder
Outcomes - Episodes of one holder kind.
- Inapplicable
Work - Work outside the active graph, with the derived reason.
- Milestone
Forecast - Sampled finish of one milestone, in elapsed hours after the clock reading.
- Next
Work Candidate - Ready task with a reproducible priority score and its explanation.
- Next
Work Query - Eligibility and scheduling inputs for a work recommendation.
- Next
Work Result - Versioned
nextresult: global ranking, then scope, then limit. - Open
Choices - Open decisions whose options decide which conditional work applies.
- Operation
- Audit envelope for one successfully applied command.
- Operation
Facts - What a committed operation records about itself, as the project store returned it.
- Outside
Scope - Eligible work that the scope excluded, kept visible rather than hidden.
- Path
Step - One activity on the mean critical path.
- Pert
Approximation Report - The analytic approximation and the path it follows.
- Progress
Projection - Derived progress values, never persisted as replacement domain objects.
- Progress
Summary - Execution progress is distinct from independent acceptance of the result. The default is no reported progress and no acceptance.
- Provisional
Release - A provisional edge released on a predecessor’s pending attempt; a start records it as basis.
- Ratio
Group - Distribution of actual/estimated ratios in one group.
- Reason
Count - How many items one reason left out.
- Reliability
- How claims ended: each claim, or handoff to a new holder, opens one episode.
- Risk
Forecast - Sampled and analytic schedule risk of the remaining work.
- Scenario
Forecast - Remaining forecast if the open decisions select these options now.
- Schedule
Projection - The authoritative remaining-work schedule projection of one snapshot at one clock reading.
- Schedule
Span - Scheduled range of one work item in elapsed hours, with whether the critical path touches it.
- Schedule
Uncertainty - Seeded Monte Carlo completion percentiles, in elapsed hours.
- Scheduled
Relation - One outstanding constraint and how tightly it binds its successor.
- Scheduled
Times - Critical-path times of one work item the active graph schedules, in elapsed hours after the evaluation time.
- Scheduled
Work - One row of the schedule projection.
- Scope
Member - A resolved scope entry, echoed so callers see exactly which entities were applied.
- Scoped
Candidate - A candidate within scope, carrying its position in the unfiltered ranking.
- Status
Summary - Execution counts and remaining schedule projections.
- Throughput
- Tasks verified in recent periods, counted by their verification time.
- Wait
Group - Distribution of elapsed waits for one actor kind.
- Wait
Report - Waits grouped by the kind of actor that ended them.
- Week
Count - Tasks verified in one ISO week.
- Work
Explanation - Execution contract, prerequisites, and derived reason for a work item.
- Work
Scope - Project and asset membership filter; an empty axis does not filter.
Enums§
- Advisory
- Advice about the caller’s situation, shown beside the ranking.
- Command
- A semantic state change; adapters must use the engine to apply it.
- Criticality
Basis - Which projection a candidate’s criticality component comes from.
- Edge
Relaxation - How a reviewed change weakens an edge that touches the applying actor’s own work.
- Engine
Error - A rejected command or failed execution projection.
- Exclusion
Reason - Why recorded work or a decision was left out of a measurement.
- Hour
Basis - How actual execution time is counted.
- Link
Refusal - Why an operation cannot be attributed to a run.
- Progress
Scope - Whether a work item’s own conditions let it count toward progress.
- Relaxed
Constraint - A constraint that a reviewed change would relax for work the applying actor owns.
- RunError
- Why a run request was refused.
- Scope
Error - A scope key that does not name an entity in the current plan.
- Transition
- Task lifecycle transition whose relation gates are evaluated.
- Unmet
Gate - Machine-readable reason preventing a lifecycle transition at the evaluated time.
Constants§
- BULK_
WINDOW_ SECONDS - Start and submit operations committed this close together, neither naming an occurrence time, were recorded after the fact in one sitting, so their interval says nothing about the work.
- MAX_
APPLIED_ RATIO - Largest group median a calibrated forecast applies.
- MAX_
SCENARIOS - Scenario forecasts are computed only up to this many option combinations.
- MIN_
ACTUAL_ SECONDS - Actual working time below this is no measurement: start and submit were recorded together, or the whole interval fell outside the task’s working hours.
- MIN_
AGENT_ APPLIED_ RATIO - Smallest group median a calibrated forecast applies to agents. Agents finishing human-scale
estimates routinely measure 0.01-0.1, so their floor sits below that range; near-zero samples
are already excluded by
MIN_ACTUAL_SECONDS. - MIN_
APPLIED_ RATIO - Smallest group median a calibrated forecast applies to people and services; lower says the records, such as times typed in afterwards, not the estimates, are wrong.
- MIN_
SAMPLES - Samples a group needs before a calibrated forecast applies its median.
- NEXT_
RESULT_ VERSION - Shape version of
NextWorkResult; incremented on any incompatible field change. - REVIEW_
FLOOR_ SECONDS - A review recorded this soon after its submission was recorded together with it; no one reads and judges a result that fast, so it is no wait.
- STALE_
AFTER - How long a working or waiting run may go unheard before it is reported stale.
- THROUGHPUT_
WEEKS - ISO weeks the throughput history covers, the current one included.
Functions§
- advisories
- Advice for an actor about to choose work: whether it already holds claims.
- apply_
command - Apply one validated semantic command atomically and return its audit operation.
- apply_
plan_ change - Apply a reviewed full proposal as one delta operation.
- authorize_
run - Refuse an agent that is not the run’s executor; humans and services may record any run.
- calibration
- Measure estimates, waits and flow from a validated plan and its operation log in append order.
- check_
link - Check that attributing
operationtorecordcannot be wrong. - check_
run_ transition - Check a move from the run’s current state.
- completion
- Derive completed task, milestone and package identities at a clock reading.
- describe_
applicability - Human-readable reason work is outside the active graph, naming decisions and work by key.
- excluded_
in_ flight - Claimed, started, blocked or submitted work outside the status scope: a reviewed choice change
excluded it, so it takes no transition until the plan changes again.
statuscounts it asexcluded_in_flight. - explain_
work - Explain a validated work item at an adapter-supplied time, including derived completion.
- gate_
report - Explain whether a transition is permitted at an adapter-supplied time, without changing state.
- in_
status_ scope - Whether work’s lifecycle counts in
statuslifecycle counts,completeand progress: its own conditions are selected. Views that list work behind those counts use this scope, so a list never shows work its count leaves out. - is_
ready - Whether a planned task can be claimed at an adapter-supplied time.
- is_
unestimated - Whether
workenters the remaining forecast at 0 h only because it has no estimate. - json_
equivalent - Whether two JSON values carry the same data. Numbers compare by value: a client that decodes and re-encodes
a reviewed change may write
2.0as2, and that is the state it was reviewed against, not a stale one. - min_
applied_ ratio - Smallest group median a calibrated forecast applies to an executor kind.
- next_
in_ scope - Rank eligible work on the full graph at an adapter-supplied time, then partition it by scope and apply the limit.
- next_
work - Recommend only tasks claimable at an adapter-supplied time and compatible with capabilities.
- observe_
run_ start - Validate a start against the plan and capture the contract it observes.
- patch
- Apply entity changes to
currentwithout any policy check. - plan_
change - Turn a reviewed full proposal into the delta command the operation log records.
- progress
- Compute execution percentages without inferring work from elapsed time or changing the schedule.
- project_
run - Compute a run’s view at
now. - propose_
change - Validate a proposed plan without changing state, lifecycle, evidence or decision outcomes.
- propose_
revert - Review the change that undoes a recorded plan change, against the current plan.
- protected_
work - Work whose authored contract and position are fixed by execution, including its ancestors and transitive prerequisites. Adapters may use these anchors without duplicating review policy.
- run_
start_ event - The transition identity of a run’s start, which is the run’s own identity.
- schedule_
projection - Project the remaining schedule;
probabilisticadds the simulationstatusandnextuse. - schedule_
span - Scheduled range of
item, orNonefor work the timeline places outside the active graph; packages span their applicable descendants. This is the rule the terminal Gantt draws. - show_
work - Return a work projection with aggregate completion reflected in its lifecycle.
- status
- Project validated lifecycle counts and remaining duration at an adapter-supplied time.
- status_
at_ horizon statuswhose forecast also states the probability of finishing the project, and of reaching each milestone, withinhorizon_hourselapsed hours ofnow.- status_
calibrated statuswhose forecasts scale each unfinished task’s estimate by its executor kind’s measured ratio and add the verifier kind’s measured review wait, each only when enough samples support it;calibrationin the result states what was applied and why. The plan itself is never changed.- status_
calibrated_ at_ horizon status_calibratedwith the horizon probabilities ofstatus_at_horizon.- unestimated
- Keys of every work item
is_unestimatedreports, in key order.