imessage-mcp

# v1.3+ Claude Plugin Expansion Research Prompt

Use this prompt in Claude Code to produce a decision-complete proposal for the next plugin phase:

```text
You are planning v1.3+ plugin expansion work for the imessage-mcp repository.

Repository context:
- Path: /Users/anipotts/Code/imessage-mcp
- Current release target: v1.2.0 credibility sprint (minimal plugin baseline)
- Existing plugin artifacts:
  - .claude-plugin/plugin.json
  - .claude-plugin/marketplace.json
  - .mcp.json
  - skills/search/SKILL.md
  - skills/wrapped/SKILL.md
  - skills/stats/SKILL.md

Research objectives:
1. Propose concrete additions for hooks, subagents, and extra plugin skills/commands.
2. Align all recommendations with official Anthropic Claude Code plugin docs.
3. Preserve strict privacy posture (read-only MCP, no accidental data exfiltration, explicit user control).
4. Produce an implementation-ready spec with no open decisions.

Constraints:
- Keep v1.3 additive; do not break existing /imessage:* commands.
- Prioritize low-risk trust features before high-risk automation.
- Any feature with side effects must be opt-in and guarded.
- Do not assume unsupported plugin schema fields.

Required output sections:
1. Proposed scope (what ships in v1.3 vs deferred)
2. Plugin file tree changes (exact files and paths)
3. Manifest/schema updates (JSON examples)
4. Hook design (events, matcher patterns, commands, safeguards)
5. Subagent design (roles, prompts, tool boundaries)
6. Additional skills/commands (names, frontmatter, behavior)
7. Privacy and safety analysis (threats, mitigations, residual risk)
8. Rollout strategy (alpha/beta/GA, migration notes)
9. Test plan (unit, integration, validation via `claude plugin validate`)
10. Acceptance criteria checklist

Deliverables must be specific enough for direct implementation by another engineer.
```