chore: track claude skills, tools, templates, reference code and research-wiki

- Add all claude skills (brainstorming, commit, debugging, TDD, etc.)
- Add claude hooks (pre-commit-guard, post-edit-quality)
- Add research templates (experiment plan, research brief, etc.)
- Add claude tools (arxiv/semantic_scholar/openalex fetch, wiki, exa)
- Add TRM4 reference implementation as algorithm fidelity baseline
- Add research-wiki content (plans, index, graph, query_pack)
- Update .gitignore to exclude .graphify_version runtime state

.claude/skills/research-wiki/SKILL.md

@@ -0,0 +1,89 @@
+---
+name: research-wiki
+description: "Manage the project research knowledge base. Initialization, querying, statistics, and health checks. Trigger phrases: knowledge base, research wiki, wiki query, check the knowledge base."
+argument-hint: [subcommand: init|query|stats|lint|ingest]
+---
+
+# Research Wiki Management
+
+Manage the project research knowledge base: $ARGUMENTS
+
+## Overview
+
+Research Wiki is the project's single source of truth for knowledge. All research and development knowledge produced by skills is stored here.
+
+### Entity Types (12)
+
+| Entity | Directory | node_id Format | Primary Source Skill |
+|---|---|---|---|
+| paper | `papers/` | `paper:<slug>` | `research-lit` |
+| idea | `ideas/` | `idea:<id>` | `idea-creator` |
+| experiment | `experiments/` | `exp:<id>` | `run-experiment` (future) |
+| claim | `claims/` | `claim:<id>` | `novelty-check` |
+| gap | `gaps/` | `gap:<id>` | `research-lit` / `idea-creator` |
+| design | `designs/` | `design:<id>` | `brainstorming` |
+| finding | `findings/` | `finding:<id>` | `systematic-debugging` / `verification-before-completion` |
+| adr | `adrs/` | `adr:<id>` | 架构决策记录 |
+| plan | `plans/` | `plan:<id>` | `writing-plans` |
+| review | `reviews/` | `review:<id>` | `requesting-code-review` / `receiving-code-review` |
+| schema | `schemas/` | `schema:<id>` | `structured-logging` |
+| metric | `metrics/` | `metric:<id>` | `structured-logging` / `harness-eval` |
+
+### Key Files
+
+| File | Purpose |
+|---|---|
+| `query_pack.md` | Compressed summary |
+| `index.md` | Category index |
+| `log.md` | Audit log |
+| `graph/edges.json` | Relationship graph |
+
+## Subcommands
+
+### `/research-wiki init`
+
+Initialize the wiki:
+
+```bash
+.claude/tools/research_wiki.py init research-wiki/
+```
+
+### `/research-wiki query "<topic>"`
+
+Rebuild `query_pack.md`:
+
+```bash
+.claude/tools/research_wiki.py rebuild_query_pack research-wiki/
+```
+
+Then show the content to the user.
+
+### `/research-wiki stats`
+
+Show statistics:
+
+```bash
+.claude/tools/research_wiki.py stats research-wiki/
+```
+
+### `/research-wiki lint`
+
+Run health checks:
+
+```bash
+.claude/tools/research_wiki.py lint research-wiki/
+```
+
+### `/research-wiki ingest "<title>" — arxiv: <id>`
+
+Import a paper:
+
+```bash
+.claude/tools/research_wiki.py ingest_paper research-wiki/ --arxiv-id <id>
+```
+
+## Key Rules
+
+- Other skills call `.claude/tools/research_wiki.py` directly; they do not go through this skill.
+- If the wiki does not exist (`research-wiki/` directory is missing), all write operations are skipped silently.
+- Failed ideas are always preserved in `query_pack`.