Capability family audit
Identify coherent family entrypoints without replacing direct building-block use.
This audit reviews the current portable skill inventory for user-facing families. A category is not a family by itself. A family is justified only when it offers a recognizable entry question, a small set of reusable building blocks, a bounded composition, and a useful gain for users or agents.
Decision vocabulary
- Established — family entrypoint exists and has a validated composition contract.
- Pilot candidate — the building blocks and entry question are sufficiently coherent for a focused contract pilot.
- Needs framing — related skills exist, but the family boundary or primary outcome is not yet clear.
- Direct use — individual building blocks are clearer than a family at current maturity.
- Hold — do not create a family until usage evidence or ownership boundaries improve.
Inventory
| Candidate family | Entry question | Building blocks | Decision | Reason |
|---|---|---|---|---|
| Product Management | “What product decision are we trying to make?” | product-management, feature-value-governance, opportunity-tree-alignment, feature-planning, observability-review, triad-check |
Established | Existing entry contract and validated modular workflow. |
| AI Discovery & Agent Experience | “What prevents this property from being discovered, represented, measured, or safely used?” | Family entrypoint plus five domain-pack specialists | Established | Existing family entrypoint and selective composition contract. |
| Work Continuity & Efficiency | “How do we make this work smaller, resumable, and safe to hand off?” | task-chunking, checkpoint-review, handoff-summary, communication |
Pilot candidate | Clear outcome and small building-block set; good next pilot. |
| Experience Design | “What experience decision needs to become clearer or safer?” | ux-strategy, ux-provocation, heuristic-audit, ux-writing, design-system-intelligence |
Needs framing | Strong members, but strategy, critique, writing, and system review are different entry questions. |
| Knowledge Governance | “Can this knowledge be trusted, reconciled, and used within its boundary?” | knowledge-source-evaluation, knowledge-conflict-resolution, restricted-context-check, domain-language-alignment |
Pilot candidate | Coherent evidence and boundary problem, but ownership and source scope need a focused pilot. |
| Engineering Delivery | “What is the smallest safe path from change intent to proven implementation?” | workflow, feature-planning, testing, debugging, lean-implementation, architecture-review, api-design, refactoring, code-style |
Needs framing | Broad lifecycle family risks becoming a second workflow runtime. Start with a narrower delivery question. |
| Business & Value | “How does this initiative create or protect value?” | business-design, revenue-lever-mapping, Product Management value skills |
Hold | Significant overlap with Product Management and unresolved decision ownership. |
| Quality & Observability | “What evidence shows the system or outcome is healthy?” | qa-review, observability-review, testing |
Hold | Quality, testing, and metrics cross several existing families. |
| Crisis Response | — | Crisis Monitor skills | Out of scope | Crisis Response belongs to Crisis Monitor and is not an Adaptive Skills family candidate. |
| Planning & Intent | “What must be clarified or stress-tested before work begins?” | intent-clarification, premortem, workflow |
Direct use | Small, cross-cutting building blocks are clearer than a new family. |
| Documentation | “How do we create a self-service document for this reader and task?” | documentation and its internal modules |
Direct use | The skill already provides the appropriate primary entrypoint. |
Recommended sequence
- Pilot Work Continuity & Efficiency with a small family entry contract.
- Pilot Knowledge Governance only after defining source ownership and restricted-context boundaries.
- Frame Experience Design around one entry question before composing its members.
- Keep Business & Value and Quality & Observability on hold until their overlaps are resolved.
- Revisit Engineering Delivery as a dedicated workflow, not automatically as a broad family.
Explicit boundary: Crisis Monitor
The Crisis Response skills are excluded from this Adaptive Skills family audit. They belong to the Crisis Monitor product and should be governed, documented, and evolved in that repository. They must not be treated as general-purpose Adaptive Skills, increase the portable-skill inventory, or shape the public Adaptive Skills navigation. Existing integration-specific files remain a separate boundary until their ownership and migration are handled explicitly.
Family creation gate
Before creating a new family, require:
- one outcome-oriented entry question;
- at least two building blocks with independent direct-use contracts;
- a selective composition path, not an all-members checklist;
- explicit exclusions and authority boundaries;
- at least one synthetic case for direct use and one for family-led composition;
- a verification result and named handoff;
- evidence that the family reduces user or agent routing ambiguity.
Until these conditions are met, keep the skills discoverable and usable directly in the catalog.