superpipelines

Audit one or all pipelines against layout, frontmatter, topology, and runtime-safety standards — produces a severity-classified report (SEV-0/1/2/3) with cited evidence

Set, change, or audit model preferences for pipeline agents — interactive 6-mode workflow (global/workspace prefs, per-agent overrides, first-run wizard, catalog refresh)

Delete a step from an existing pipeline — select pipeline, select step, perform gap analysis, optionally rewire, audit the delta, then gate on human approval before any deletion

Initiates a repository traversal to create localized PIPELINE-CONTEXT.md hierarchical context files.

Migrate an existing pre-v2 per-tier pipeline (.claude/, .opencode/, .agents/codex/) into the unified data-only .superpipelines/ layout — select legacy pipeline, translate frontmatter to canonical agent defs, stage, delta-audit, gate on human approval, then atomically promote, rewrite the registry, and move the legacy artifacts

Use when designing a new multi-agent pipeline, generating spec/plan/tasks/topology artifacts, adding a step to an existing pipeline, updating a step, deleting a step, creating a single subagent definition, or diagnosing a pipeline topology failure.

Use when auditing existing pipeline bundles, agent files, or skills against superpipelines v2 layout, frontmatter, topology, and runtime-safety standards. Invoked automatically after new-pipeline, new-step, update-step, and delete-step. Produces severity-classified reports (SEV-0/1/2/3) with cited file:line evidence.

Use during a Pattern 3 iterative loop after a tester reports failures, before dispatching a fixer — diagnoses whether failures are fixable bugs or architectural problems, detects "fixes reveal new failures in new locations" pattern, and decides whether to continue or escalate per Pattern 3 protocol.

Use when surveying an existing pipeline bundle for optimization opportunities — topology merge/split/parallelize/reorder, model-tier right-sizing, past-run pain signals, and protocol-quality advisories. Dispatched read-only by optimizing-a-pipeline; renders an opportunity report for the orchestrator to persist. Never mutates.

Use as Stage 2 review ONLY after pipeline-spec-reviewer returned PASS — checks code quality, idiom, maintainability, naming, structure, and tests against the spec. Refuses to run if Stage 1 not yet PASSed. Read-only; never edits.

Supplies the pipeline-auditor agent with its operating modes (FULL/DELTA/SCOPE-WIDE), full compliance matrix dispatch, severity classification, fix routing, and model-tier resolution criteria. Use when the pipeline-auditor agent is invoked to perform standards-enforcement auditing of pipeline bundles.

Supplies the iteration-control protocol, escalation hard-gate, and rationalization resistance for Pattern 3 diagnosis cycles. Use when the pipeline-failure-analyzer agent is invoked and requires its operational protocol, convergence detection logic, or escalation criteria.

Full operational protocol for the pipeline-optimizer agent — read-only four-axis optimization survey.

Supplies the pipeline-quality-reviewer agent with the Stage 2 operational protocol, finding classification rules, verdict logic, and rationalization resistance tables for code quality audits. Use when the pipeline-quality-reviewer agent is dispatched to perform Stage 2 review after Stage 1 has returned an explicit PASS verdict.

Supplies the Stage 1 functional-verification protocol to the pipeline-spec-reviewer agent: acceptance-criteria compliance checks, under-build/over-build detection, binary verdict emission, and rationalization resistance. Use when the pipeline-spec-reviewer agent activates to verify that an executor's output matches its specification before Stage 2 may proceed.

1 hook across 1 event