What problem does it solve? Autonomous agent pipelines often become fragmented collections of orphan scripts with hidden dependencies and no verifiable execution proof. This Skill enforces a single explicit graph SSOT where every node has a command vector, receipt file, typed edges, and tier placement, so multi-agent systems stay governable and auditable. ## Core Features & Use Cases - Explicit node graph design: Define nodes with snake_case ids, command vectors, receipt paths, planes (INTERNAL, RUNTIME_GLUE, HUB_API, PORTFOLIO), and typed edges (trigger, gate, wire, fan-out, data, orchestrate, surface, glue, alarm) in a versioned JSON graph. - Parallel tier orchestration: Organize nodes into T0 Safety, T1 Truth, T2 Fleet, and T3 Proof tiers that run sequentially across tiers and in parallel within tiers via a ThreadPool runner. - Change classification and proof ladder: Classify changes as G0–G3 and apply the matching validation, from graph linting to full runner execution with JSON receipts. - Use Case: When adding a new live-wire validator to an agentic system, follow Pattern A to create the script, register the graph node with edges_in/edges_out, update the directory map and mesh catalog, then run validate-pipeline-node-graph-v1.sh and the tier runner to produce a receipt. ## Quick Start Ask the agent to design and register a new node in the pipeline graph, then validate it with the graph validator and run its tier to produce a receipt.
