Skip to content

Composition runbook

This page is the human-readable version of the composition chains. The machine-readable per-pattern composition manifests are at vstack://patterns/<name>/composition via the MCP server, or GET /v1/patterns/<name>/composition via the REST API.

For the full curated runbook covering all five canonical chains (failure / team / structural / culture / calibration), see COMPOSITION-RUNBOOK.md at the repo root.

The five canonical chains

Chain Trigger Patterns Skill
F1 — Confidently wrong Agent returned wrong answer with high confidence AAR → Lewin → branch on locus (Bias Stack / Yerkes-Dodson / Glaser) /vstack-post-incident
T1 — Crew is "off" Multi-agent crew produces output that doesn't feel right Lencioni + Edmondson + Trust Triangle + Process Gain/Loss + Bias Stack (parallel) /vstack-audit-crew
S1 — Bottleneck under load Crew slows down or backs up under traffic Span-of-Control (deterministic) → Org-Structure (qualitative) → Social Loafing / Superflocks (behavioral) /vstack-bottleneck
C1 — Culture drift Crew's behavior doesn't match the team's intent Schein iceberg → Robbins-Judge 7-characteristic → (optional) McGregor /vstack-culture-check
D1 — Calibration Set + verify monitoring baselines for drift detection Any 1-N patterns in forensic mode, baseline JSON written to ~/.vstack/baselines/ /vstack-baseline

Cross-chain transitions

The most common multi-chain transitions (from the runbook):

If this chain... ...surfaced this ...the natural next chain is
F1 (failure) Lewin says interactional T1 (full crew audit)
F1 Lewin says environmental + crew is multi-agent S1 (bottleneck)
T1 (team) Lencioni "absence of trust" C1 (culture)
T1 Lencioni "lack of commitment" S1 (bottleneck)
S1 (structural) Math fine + structure wrong C1 (culture root often masquerades as structure)
C1 (culture) Schein layer-drift severity high F1 against a specific failed run (concretize the drift)
Any Drift suspected over time D1 (baselines) — then re-run quarterly

The output shape for the executive readout

Every chain ends with one structured readout. The shared template:

## <Chain name> — <one-line scope>

**Headline:** <one sentence — deepest finding with severity>

**Layered view:**
- <pattern 1>: <severity> + <one-line top finding>
- <pattern 2>: <severity> + <one-line top finding>
- ...

**The chain:** <one sentence connecting the patterns at different resolutions>

**Three highest-leverage interventions:** (deduped, ranked by estimated_impact)
1. <intervention> (from <pattern>)
2. ...
3. ...

**Where to look next:** <recommend the next /vstack-* skill>

Cap at ~500 words. Detection JSONs go in a collapsible appendix.

See COMPOSITION-RUNBOOK.md for code + per-chain detailed walkthroughs.