# MORPH-SPEC Skills

Skills installed to `.claude/skills/` during `morph-spec init` (or `morph-spec update`).

Each skill is a directory: `{name}/SKILL.md` + optional `scripts/`, `references/`, `assets/`.

---

## core/ — NLH Skills (Single Source of Truth)

These are the only skills the harness needs. MORPH.md references them by path.

| Skill                | Purpose                                                              | User-invocable |
|----------------------|----------------------------------------------------------------------|----------------|
| `morph-brainstorming` | Business understanding + brownfield scan → proposal.md (Gate 1)     | yes            |
| `morph-stack-scan`    | Detect stack drift in existing projects, report in stack-scan.md     | yes            |
| `morph-plan`          | Produce spec.md + mandate.md + tasks.json (Gate 2)                   | yes            |
| `morph-implement`     | Execute each task with standard + persona, auto-advance between tasks| no             |
| `morph-eval`          | Score work 0-10 across 4 dimensions, silence on score >= 9           | no             |
| `morph-review`        | Final review report with scores + recommendation (Gate 3)            | yes            |
| `morph-standards`     | Locate and load a specific standard on demand                        | no             |

The legacy `level-0-meta/` and `level-1-workflows/` skill trees were removed
in v5 — their useful content was consolidated into `core/`. MORPH.md is the
single charter that ties everything together.

---

## Domain Agents

Domain specialist agents (`framework/agents/`) are installed separately as
`.claude/agents/morph-{name}.md` — they are subagents dispatched via the
`Agent` tool, not skills.

See `framework/agents.json` for the full list and persona definitions.

---

*MORPH-SPEC by Polymorphism Tech*