execution-graph

name	execution-graph
description	Use when a WCBS plan, remediation backlog, migration, release, or multi-slice engineering objective has dependent work that requires requirement-to-node traceability, topology validation, READY-frontier scheduling, or evidence-backed node locking. Also activate when the user explicitly requests /graph execution.
activation	Activate for a WCBS-governed agent after initialization and APIVR tier selection when approved work has two or more dependency-bearing execution or verification slices; do not activate for one narrow independent task.
required_inputs	Current WCBS authority and authorization context, APIVR tier, applicable Elite Build Goals, approved objective or plan revision and source baseline, requirement sources, acceptance criteria, dependencies, preserved behavior, verification methods, and existing evidence or progress records.
required_outputs	A dependency-aware execution graph, requirement-to-node traceability, topology receipt and verdict, current READY frontier, evidence-backed lock decisions, graph status summary, WCBS verdict, and one next action.
authority_dependencies	00_start_here/SOURCE_OF_TRUTH.md; 00_start_here/LOAD_ORDER.md; 00_start_here/capability-routing.json; 10_governance/APIVR_EXECUTION_LIFECYCLE.md; 10_governance/RELEASE_GATES.md; 20_skills/PORTABLE_SKILL_CONTRACT.md; skills/writing-plans/SKILL.md; skills/executing-plans/SKILL.md.
evidence_requirements	Use only WCBS evidence states and records; every topology, readiness, and lock claim must cite current source, dependency, verification, or ledger evidence, and stale evidence invalidates affected locks; otherwise mark the claim Unknown, Not Run, or Blocked.

Execution Graph

Derive a deterministic scheduling view from an approved WCBS plan: every requirement is mapped, topology is valid, and only proven-ready work advances.

This is a capability inside WCBS, not a second operating system. It does not own initialization, goals, APIVR, Elite Build Goals, evidence vocabulary, specialist roles, delegation, retries, release gates, or final verdicts. Load their canonical WCBS owners instead of copying their rules here. It never expands user authorization, approved scope, or write permissions.

Scope

This skill owns only:

• conversion of approved requirements or plan slices into execution nodes;
• requirement-to-node traceability;
• dependency-edge and topology validation;
• computation and ranking of the READY frontier;
• evidence-backed node lock decisions; and
• recalculation of downstream readiness after each material state change.

This skill creates no new .wcbs authority file or parallel ledger. Place the graph in the current implementation plan or run-control artifact. Record execution and evidence in artifacts selected by LOAD_ORDER.md. A new project-scoped graph file is non-authoritative until the Controller contract declares it.

Activation Decision

Activate for prerequisites, requirements spanning slices, fan-out, or partial blockers that must not freeze independent work. An explicit /graph applies only when delivered as text with an approved objective; it proves neither native slash-command support nor implementation authority. Use executing-plans when graph computation would change nothing.

WCBS Ownership And Input Gate

Concern	Canonical owner
Authority, tier, lifecycle, verdict	source of truth, tier router, APIVR
Plan and requirement slices	writing-plans, product-requirements-and-issue-slicing
Sequential or delegated execution	executing-plans; dispatch and subagent skills
Repeated work and checkpoints	loop, long-horizon, and run-tracing skills
Verification and release	verification and release-readiness skills

Confirm initialization, tier, applicable Elite Build Goals, scope, authorization, rollback boundary, baseline, requirements, criteria, and evidence locations. Missing inputs route to canonical discovery or planning; never fabricate them. The graph inherits the tier and may trigger canonical escalation, but cannot lower it. Before declaring VALID or READY, require SOURCE BASELINE, PLAN REVISION, APIVR TIER, and AUTHORIZED SCOPE; if any is missing, set BLOCKED and leave the frontier empty. A provisional topology may diagnose the gap but cannot authorize execution.