fix: restore consensus-engineer skill deleted in v0.8.0

skills/consensus-engineer/README.md

@@ -0,0 +1,58 @@
+# consensus-engineer
+
+AI solution architect for [consensus-tools](https://github.com/consensus-tools/consensus-tools) — the decision infrastructure for autonomous AI agents.
+
+## What it does
+
+`/consensus-engineer` is an interactive skill that walks you through integrating consensus-tools into your project:
+
+1. **Analyzes your project** — detects your stack, framework, AI SDKs
+2. **Discovers your use case** — maps your needs to guard domains + consensus policies
+3. **Recommends architecture** — guards vs wrapper vs hybrid pattern
+4. **Scaffolds setup** — installs packages, creates config, generates starter code
+5. **Proves it works** — runs a live guard evaluation against sample input
+6. **Extends** — LangChain tools, AI SDK middleware, custom templates
+
+## Quick start
+
+```bash
+# In Claude Code:
+/consensus-engineer
+```
+
+The skill guides you through everything. No prior knowledge of consensus-tools required.
+
+## What's included
+
+- **llms.txt** (2,200+ lines) — Complete system reference covering all 32 packages, 29 MCP tools, 9 consensus policies, 7 guard domains, schema types, examples, and integration patterns
+- **SKILL.md** — Interactive 6-phase guided experience with AskUserQuestion gates
+- **metadata.json** — Skill metadata for ClawHub/skills.sh publishing
+
+## Consensus-tools packages
+
+The skill helps you integrate these packages:
+
+| Package | What it does |
+|---------|-------------|
+| `@consensus-tools/guards` | 7 guard domains + `createGuardTemplate()` for custom domains |
+| `@consensus-tools/policies` | 9 consensus algorithms + `createPolicyTemplate()` for custom policies |
+| `@consensus-tools/wrapper` | Runtime function gating + `createWrapperTemplate()` |
+| `@consensus-tools/personas` | Persona packs, reputation engine, respawn logic |
+| `@consensus-tools/langchain` | Guards as LangChain tools + LangSmith tracer |
+| `@consensus-tools/ai-sdk` | Guarded generateText/streamText for Vercel AI SDK |
+| `@consensus-tools/core` | Job engine, ledger, vote aggregation |
+| `@consensus-tools/schemas` | Zod schemas + TypeScript types |
+| `@consensus-tools/mcp` | 29 MCP tools for Claude integration |
+
+## Integration patterns
+
+The skill detects which pattern fits your project:
+
+- **Guards** — API/workflow style: evaluate inputs before actions execute
+- **Wrapper** — In-memory function gating: wrap functions with reviewer consensus
+- **Hybrid** — Guard templates as wrapper reviewers: best of both worlds
+- **MCP** — 29 Claude-native tools for AI-driven governance
+
+## License
+
+Skill: MIT | Engine: Apache-2.0