Work Slice Visual State Contract

Posture: docs_first, read_only_projection. This contract defines presentation lanes and a derived card shape. It does not define a Work Slice lifecycle or state machine.

Purpose

Define how a cockpit may summarize one governed Work Slice without replacing its task, decision, execution, readiness, handoff, restart, telemetry, or learning records.

Non-goals

no new canonical Work Slice schema;
no automatic transition or board mutation semantics;
no replacement for readiness outcomes;
no requirement that every slice pass through every lane;
no inference of success from activity, token usage, or a runtime completion event.

Presentation lanes

intake
framing
context_ready
planning
execution
validation
human_review
reconcile
closed
blocked

These lanes are a visual grouping vocabulary. A slice can skip lanes, return to an earlier lane, or remain outside the board if the source artifacts do not support a useful projection.

Derived state shape

work_slice_visual_state:
  work_slice_id:
  title:
  objective:
  presentation_lane:
  lane_confidence: confirmed | inferred | conflicted | unknown
  risk_level: low | medium | high | critical | unknown
  planning_depth: lite | standard | high_assurance | unknown
  readiness_outcome: continue | tighten | review | handoff | escalate | stop | unknown
  evidence_status: none | partial | sufficient | failed | inconclusive | unknown
  human_review:
    required: true | false | unknown
    status: not_requested | pending | completed | unavailable
    reviewer_role:
    open_question:
    source_refs: []
  primary_skill:
    skill_id:
    activation_status: active | completed | unavailable | unknown
    source_refs: []
  runtime:
    runtime_id:
    session_status: active | completed | failed | stopped | unavailable | unknown
    source_refs: []
  telemetry:
    tokens:
      value:
      provenance: reported | estimated | unavailable
      source_refs: []
    cost:
      value:
      currency:
      provenance: reported | estimated | unavailable
      source_refs: []
  alerts: []
  evidence: []
  source_refs: []
  projected_at:

Every entry in alerts and evidence must carry its own source_refs.

Lane derivation guidance

Lane	Minimum supporting signal
`intake`	A Work Slice or candidate exists but framing is incomplete
`framing`	Objective, scope, risk, or non-goals are being made explicit
`context_ready`	Context minimum is recorded as adequate for the next step
`planning`	A plan, spec bundle, pattern selection, or readiness preparation is active
`execution`	An authoritative execution record or runtime session is active
`validation`	Execution output exists and required evidence is being checked
`human_review`	An authoritative source requires or requests human review
`reconcile`	Closeout, learning, documentation, or continuity reconciliation is active
`closed`	The authoritative slice record reports closure with required evidence
`blocked`	A gate, policy, missing authority, failure, or stop condition prevents progress

Activity alone is insufficient. For example, runtime.session_completed does not derive closed; the evidence and authoritative closure record must support that lane.

Precedence and conflicts

Explicit project or AletheIA records outrank a lane inferred from low-level runtime activity.
A current stop, unresolved block, or pending mandatory human review prevents a confident closed lane.
Conflicting authoritative records produce lane_confidence: conflicted plus an alert containing all relevant source_refs.
If no rule is sufficiently supported, use presentation_lane: intake only when a candidate record exists; otherwise use unknown in projections that permit it.

Evidence posture

none: no evidence references exist;
partial: some expected evidence exists, but the source requirements are incomplete;
sufficient: the authoritative validation or review record says the evidence is enough;
failed: required evidence records a failure;
inconclusive: evidence conflicts or cannot establish the claim;
unknown: the projection cannot determine the posture.

The cockpit does not promote partial to sufficient by counting artifacts.

Alerts

A derived alert must contain:

alert:
  alert_id:
  type:
  severity: info | warning | critical
  summary:
  suggested_review:
  source_refs: []
  resolved_at:
  resolution_ref:

Alerts request attention. They do not authorize, deny, close, or reopen work.