Skills Ecosystem
Build reusable agent skills, commands, and documents for repeated engineering workflows.
A skills ecosystem turns repeated work into reusable instructions. It gives agents and humans a shared way to perform repository-specific tasks without rediscovering the same process.
Skill Candidates
| Workflow | Skill value |
|---|---|
| New app setup | Repeat workspace, config, deploy, and verification steps |
| UI component docs | Preserve MDX and registry structure |
| SEO audit | Standardize page, metadata, and indexing checks |
| SDK installation | Apply package-specific setup safely |
| Release verification | Run the same pre-release checklist |
| Incident triage | Collect logs, affected services, and rollback options |
Skill Design Rules
- Write for a concrete workflow, not a broad topic.
- Include trigger phrases so the skill is used at the right time.
- Prefer scripts or templates for mechanical steps.
- Keep instructions current with repository structure.
- Record verification commands and expected outputs.
Ecosystem Governance
| Item | Owner decision |
|---|---|
| Skill owner | Who updates the instruction |
| Review cadence | When the skill is tested or revised |
| Scope | Which apps or packages it may affect |
| Safety | What actions require human approval |
Quality Signal
A skill is useful when it reduces repeated explanation, prevents common mistakes, and still leaves room for engineering judgment.