# Evozeus Wrapper Target Skill Diagnosis

> Use when identifying an independent Skillware Git Repo, its local installs, GitHub state, root Harness state, architecture, evolution surface, component gaps, and publication boundary.

- Skill: `metainflow/evozeus-wrapper-target-skill-diagnosis` (Agent Skill)
- Install (CLI): `npx get-skill add metainflow/evozeus-wrapper-target-skill-diagnosis`
- Raw SKILL.md: https://api.skillmd.com/api/skills/metainflow/evozeus-wrapper-target-skill-diagnosis/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Web & Frontend
- Author: metainflow (https://skillmd.com/u/metainflow)
- Updated: 2026-11-16
- Page: https://skillmd.com/skills/metainflow/evozeus-wrapper-target-skill-diagnosis

---


# Target Skill Diagnosis

Use this stage after environment diagnosis and before attach, adopt, repair, publish, or reinstall.

## Required Inputs

- Target path inside an independent Git Repo. The CLI normalizes it to the Repo root; Harness ownership always stays at that root.
- Target GitHub repo in `OWNER/REPO` format when known.
- Skill name override only when the target surface cannot identify it.

## Command

```bash
python3 scripts/evozeus_wrapper.py skill diagnose \
  --target /absolute/path/to/skill \
  --repo OWNER/REPO \
  --json
```

## Decide

Ask the user only when:

- The target path is missing or has no detectable evolution surface.
- Multiple local repos can be the canonical source.
- Multiple install copies differ from the canonical repo.
- Visibility is not explicitly `public` or `private`.
- Private data could enter docs, Issues, release notes, or Pages.

## Required Order

1. Confirm environment diagnosis has passed. If `~/.evozeus` is missing, install / initialize EvoZeus before target transform.
2. Confirm the target resolves to an independent Git Repo root and no nested Harness manifest exists.
3. Check GitHub Repo access, visibility, default branch, and current account permission. Record whether `ADMIN` authority is available for later writes.
4. Classify the target architecture:
   - `single_skill`
   - `runtime_skill_bundle`
   - `hooked_skill_bundle`
   - `skill_bundle`
   - `agents_runtime`
   - `unknown`
5. Report runtime integration mode and each scoped capability from `skill.integration`. Distinguish project maintenance, user-level global dispatch, Skill-entry preflight, tool gateway, and future Skill invocation hooks. Do not call project hooks or wrapper CLI commands native per-Skill invocation hooks.
6. Report Skill inventory from `skills/*/SKILL.md` when present.
7. Report `evolution_surface` facts: candidate instruction surfaces, controller files, and evidence boundaries. Do not treat script candidates as final placement.
8. Use `skills/evolution-surface-diagnosis/SKILL.md` to browse the whole repo and choose the controlling instruction surface.
9. Report `component_gaps`: missing wrapper files, manifest, changelog, and status-check concept after the surface decision is known.
10. Hand the diagnosis JSON and surface decision to `skills/status-assessment/SKILL.md`; do not write user-facing status analysis in the script.

## Stop Conditions

- Target Skill identity is ambiguous.
- Target is outside Git or contains an active nested Harness.
- The user has not chosen visibility.
- Sensitive data cannot be safely redacted.

