task-dag-decomposition

FoundationJIT only

Analyze candidate task nodes, dependencies, collisions, and critical path before a consumer accepts or rejects a final DAG.

Use when

Trigger signals that make this skill the right owner

candidate work units need evidence-backed analysis of nodes dependency edges collisions cycles critical path or parallel opportunity before a consumer accepts a final graph

Anti-triggers — as binding as the triggers above

one acceptance outcome and produced output cannot be separated into useful candidate nodes
an authoritative graph has already been accepted

What must be supplied before this skill can decide anything

acceptance outcomes and candidate produced outputs
current evidence for producer-consumer relationships contracts and order constraints
candidate read write shared-resource and collision surfaces
known uncertainty about nodes edges ownership or resources

What this skill owes the next role

candidate-graph evidence with acceptance-linked nodes, produced outputs, evidence-backed data/control/contract/order edges, rejected edges, collision/shared-write/resource surfaces, candidate critical path, parallel opportunity, cycles, uncertainty, proof limits, and residual risk for consumer acceptance or rejection

When this skill must hand the decision back

candidate output ownership edge evidence collision identity or cycle resolution remains uncertain

This skill never owns a task. These professionals may pull it in as Layer 3.

Each reference carries its own load condition. None of them enter context by default.

Reference	Load when	Do not load when	Required by
candidate-graph-evidence.mdevidence-pattern	candidate nodes edges collisions cycles critical path or parallel opportunity need source-backed comparison	one candidate graph has accepted node-output bindings and every edge is supported or rejected	Analysis