Skip to content
Back to search

ai-research-skills

Researcher-facing Claude Code skills for Zotero, Obsidian, NotebookLM, research memory, and paper workflows.

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today

Check it deeper on your machine

It has changed since its scan on 13 Nov 2026.

As a Claude Code plugin, install it and then ask Claude Code:

/plugin marketplace add keithmackay/mackayi/plugin install scanskillsafety@mackayiDo a deep scan of https://github.com/WenyuChiou/ai-research-skills with scanskillsafety

Or from a terminal, clone it and run it:

git clone https://github.com/keithmackay/scanskillsafety-skill ~/.claude/skills/scanskillsafetynode ~/.claude/skills/scanskillsafety/cli.cjs --deep 'https://github.com/WenyuChiou/ai-research-skills'

It checks the exact commit: every file, its packages against OSV (api.osv.dev), how it compares with this page's scan (findsafeskills.com receives the repository's address), and, if you agree, Claude reading every file with your own Claude subscription. It never runs the code.

Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/wenyuchiou--ai-research-skills/badge.svg)](/listings/wenyuchiou--ai-research-skills)

HTML

<a href="/listings/wenyuchiou--ai-research-skills"><img src="/api/listings/wenyuchiou--ai-research-skills/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

The core asset is portable `SKILL.md`; Claude Code marketplace support is the convenience layer. Use the path that matches your host:

| Host | Path | |---|---| | Claude Code | Marketplace install below; gets auto-triggering and `claude plugin list` verification. | | Codex CLI / Cursor / Windsurf | Load the canonical repo's `SKILL.md` into the host's skills/rules directory, or inline it into the prompt. | | Gemini CLI / generic API client | Inline `SKILL.md` as prompt or system context. | | Hermes | Install a raw `SKILL.md`; this repo has Hermes 0.13.0 skill-load verification for `literature-triage-matrix`. | | OpenClaw | Use a `SKILL.md` directory shape such as `~/.openclaw/skills/<skill>/SKILL.md` when supported; not release-verified here yet. |

Full per-host recipes are in [docs/install.md → Using these skills outside Claude Code](docs/install.md#using-these-skills-outside-claude-code).

**Claude Code fastest path — 30 seconds:**

claude plugin marketplace add WenyuChiou/ai-research-skills
claude plugin install research-workspace@ai-research-skills

**Windows `cmd.exe`:** run the commands one at a time. If the plugin list does not show `research-workspace@ai-research-skills`, your paste probably only executed the first line; see [docs/install.md](docs/install.md) for the single-line CMD form.

For Claude Code, the first two commands install the 12-skill `research-workspace` plugin. Add the optional plugins only when you need manuscript writing, full Zotero operations, or Claude-to-Codex/Antigravity delegation.

**Additive install — stop after any step and use what you have:**

# 1. Marketplace + 12 research-hub skills
claude plugin marketplace add WenyuChiou/ai-research-skills
claude plugin install research-workspace@ai-research-skills

# 2. Manuscript work — outline, drafting, prose and evidence audit, scientific review (14/17)
claude plugin install academic-writing-skills@ai-research-skills

# 3. Zotero CRUD (15/17; enable local API in Zotero desktop first — docs/setup-guide.md §C)
claude plugin install zotero-skills@ai-research-skills

# 4. Multi-CLI delegation (17/17; install codex / agy CLI binaries first)
claude plugin install codex-delegate@ai-research-skills
claude plugin install antigravity-delegate@ai-research-skills

# 5. Literature pipeline automation (Python CLI behind research-hub skills)
pip install research-hub-pipeline
research-hub setup --persona researcher
research-hub doctor

**Separate live smoke (not installation):**

This is a **live** literature workflow, not an offline install check. `--no-nlm` disables NotebookLM only; it does not disable other network calls, models, or writes. Complete the [live-run preflight](docs/live-run-preflight.md) and obtain the applicable data/cost/write authorization before running:

research-hub auto "agent-based modeling" --max-papers 3 --no-nlm

Batch all 5 plugins in one go:

bash scripts/install-all.sh        # macOS / Linux / git-bash
pwsh scripts/install-all.ps1       # Windows PowerShell

**Verify:**

claude plugin list
# fastest path: research-workspace@ai-research-skills is ✔ enabled.
# full install: 5 plugins ending in @ai-research-skills are ✔ enabled.

`claude plugin list` verifies only the Claude Code marketplace install. It does not tell you whether Codex, Cursor, OpenClaw, Hermes, or a generic API client has loaded the `SKILL.md` files. Marketplace-installed plugins live under `~/.claude/plugins/cache/...`, not `~/.claude/skills/`.

Per-plugin details: [docs/install.md](docs/install.md). Python / Zotero / Git not set up yet? Start with [docs/setup-guide.md](docs/setup-guide.md). Need to know whether a skill is prompt-only or needs the Python runtime? See [docs/runtime-contract.md](docs/runtime-contract.md).

<details> <summary><b>I'd rather clone the repo</b> (contributors / debugging)</summary>

git clone https://github.com/WenyuChiou/ai-research-skills.git
cd ai-research-skills

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

.claude-plugin/marketplace.json · Manifest · 79 lines
  1. {
  2. "$schema": "https://anthropic.com/claude-code/marketplace.schema.json",
  3. "name": "ai-research-skills",
  4. "description": "Researcher-facing Claude Code skills for Zotero, Obsidian, NotebookLM, research memory, and paper workflows.",
  5. "owner": {
  6. "name": "Wenyu Chiou"
  7. },
  8. "metadata": {
  9. "description": "Catalog of AI skills for the research workflow — researcher-led, AI-assisted",
  10. "version": "1.7.21"
  11. },
  12. "plugins": [
  13. {
  14. "name": "research-workspace",
  15. "description": "12 research-hub skills: a resumable human-in-the-loop workflow orchestrator plus literature search, comparison, orientation, context compression, design, verified multi-agent routing, NotebookLM verification, paper memory and summaries, Zotero curation, and gap-to-topic decisions.",
  16. "category": "productivity",
  17. "homepage": "https://github.com/WenyuChiou/research-hub",
  18. "version": "0.5.28",
  19. "strict": true,
  20. "source": {
  21. "source": "url",
  22. "url": "https://github.com/WenyuChiou/research-hub.git",
  23. "ref": "master"
  24. }
  25. },
  26. {
  27. "name": "academic-writing-skills",
  28. "description": "Lifecycle-aware manuscript development and general scientific review with progressively loaded display provenance, psychometrics, AI/LLM, water, flood, round, and optional reviewer modules.",
  29. "category": "productivity",
  30. "homepage": "https://github.com/WenyuChiou/academic-writing-skills",
  31. "version": "1.4.3",
  32. "strict": true,
  33. "source": {
  34. "source": "url",
  35. "url": "https://github.com/WenyuChiou/academic-writing-skills.git",
  36. "ref": "main"
  37. }
  38. },
  39. {
  40. "name": "zotero-skills",
  41. "description": "Full Zotero CRUD: search, add, update, tag, trash/restore, delete, merge duplicates, and attach PDFs via pyzotero. Dual local/Web API routing; writes work even when the separately configured zotero MCP server is read-only.",
  42. "category": "productivity",
  43. "homepage": "https://github.com/WenyuChiou/zotero-skills",
  44. "version": "0.4.0",
  45. "strict": true,
  46. "source": {
  47. "source": "url",
  48. "url": "https://github.com/WenyuChiou/zotero-skills.git",
  49. "ref": "master"
  50. }
  51. },
  52. {
  53. "name": "codex-delegate",
  54. "description": "Minimal synchronous Codex CLI adapter for legacy on-disk briefs, result/sentinel contracts and Bash/PowerShell portability; supervisor scope/test acceptance remains required.",
  55. "category": "productivity",
  56. "homepage": "https://github.com/WenyuChiou/codex-delegate",
  57. "version": "0.1.5",
  58. "strict": true,
  59. "source": {
  60. "source": "url",
  61. "url": "https://github.com/WenyuChiou/codex-delegate.git",
  62. "ref": "master"
  63. }
  64. },
  65. {
  66. "name": "antigravity-delegate",
  67. "description": "Bounded cheap-executor lane for non-honesty-critical mechanical edits through Google Antigravity CLI, with brief-first scope and result verification.",
  68. "category": "productivity",
  69. "homepage": "https://github.com/WenyuChiou/antigravity-delegate",
  70. "version": "0.1.3",
  71. "strict": true,
  72. "source": {
  73. "source": "url",
  74. "url": "https://github.com/WenyuChiou/antigravity-delegate.git",
  75. "ref": "master"
  76. }
  77. }
  78. ]
  79. }
README.md · README · 536 lines
  1. # AI Research Skills
  2. > **The catalog, installation, and visual layer for a Vertical AI Research Reference Harness.**
  3. > Move from research orientation to release through 17 portable skills,
  4. > an eight-stage lifecycle, evidence-aware handoffs, and explicit human gates.
  5. Built for graduate students, PhDs, postdocs, research support staff, and
  6. agent/harness builders who need inspectable research workflows rather than a black box.
  7. The portable skills live in this repository; `research-hub` supplies the executable
  8. runtime, while the optional `agent-collab-harness` adds policy/checkpoint governance.
  9. Languages: [English](README.md) | [繁中](README.zh-TW.md) ·
  10. [Pipeline](docs/pipeline.md) ·
  11. [Examples](docs/examples.md) ·
  12. [Runtime contract](docs/runtime-contract.md) ·
  13. [Glossary](docs/glossary.md) ·
  14. [For Agent/Harness Builders](docs/for-agent-harness-builders.md)
  15. <sub><a href="LICENSE"><img alt="License: MIT" src="https://img.shields.io/badge/license-MIT-blue"></a></sub>
  16. > 📚 Part of the [agentic AI learning roadmap](https://github.com/WenyuChiou/awesome-agentic-ai-zh)
  17. > — featured in §13–14 (research workflows).
  18. ---
  19. ## Start here — Why, what, and how
  20. ### See the whole research lifecycle first
  21. [![Seventeen research skills mapped to an eight-stage workflow from literature discovery to manuscript release](docs/img/pipeline-overview.png)](docs/pipeline.md)
  22. <sub>The 17-skill core is organized around durable handoffs: each stage produces the next stage's input, while cross-cutting tools preserve context, verification, and delegation boundaries.</sub>
  23. | Your first question | Short answer | Go deeper |
  24. |---|---|---|
  25. | **Why use these skills?** | Keep research context durable, verify evidence before claims move downstream, and pause automation at meaningful human decisions. | [Problems this catalog solves](#2-why-this-catalog-exists) |
  26. | **What is included?** | 17 portable skills, 8 lifecycle stages, 5 installable plugins, an executable research runtime, and an optional policy/checkpoint harness. | [Skill outputs](#5-see-what-each-skill-produces) |
  27. | **How do I use it?** | Install the 12-skill research workspace, then ask for a concrete research outcome in natural language. Add writing, Zotero, or delegation plugins only when needed. | [30-second install](#1-install--get-the-skills) · [Usage patterns](#4-use-it) |
  28. | 17 core skills | 8 research stages | 5 optional-by-need plugins | 3 truth-store integrations |
  29. |---:|---:|---:|---:|
  30. | Portable `SKILL.md` contracts | Orientation → release | Start with 1; grow to 5 | Zotero · Obsidian · NotebookLM |
  31. ### Try the smallest useful path
  32. ```bash
  33. claude plugin marketplace add WenyuChiou/ai-research-skills
  34. claude plugin install research-workspace@ai-research-skills
  35. ```
  36. Then ask, for example:
  37. > Build a literature triage matrix for human-in-the-loop research agents. Verify identifiers, preserve contradictions, and show me the evidence gaps before recommending next steps.
  38. Need an executable, resumable workflow instead of prompt-only skills? Continue
  39. with `pip install research-hub-pipeline`, then run `research-hub setup` and
  40. `research-hub doctor`. The full path is in [Install](#1-install--get-the-skills).
  41. ### Understand the two control layers
  42. #### Harness architecture — deterministic control around semantic work
  43. [![Vertical AI Research Reference Harness architecture showing users, research-hub runtime, role agents, evidence packets, truth stores, human gates, and the optional agent-collab policy layer](docs/img/harness-architecture.png)](docs/for-agent-harness-builders.md)
  44. <sub>`research-hub` owns the research-domain runtime and recoverable state. Role agents do semantic work; deterministic validators, evidence contracts, and the optional `agent-collab-harness` policy/checkpoint layer control what may proceed.</sub>
  45. #### Human-in-the-loop — where automation must pause
  46. [![Human-in-the-loop workflow state machine showing automatic reversible work, approval gates, decline and revise paths, recovery, and release authorization](docs/img/hitl-state-machine.png)](docs/for-agent-harness-builders.md)
  47. <sub>Read-only and reversible steps can continue automatically. Breaking contracts, skill removal, external writes, semantic acceptance, and release authorization remain explicit human decisions.</sub>
  48. | If you are… | Best next page |
  49. |---|---|
  50. | A researcher or research team | Follow the [eight-stage pipeline](#3-the-pipeline--what-each-stage-delivers-to-the-next) after the quick install above. |
  51. | An agent or harness builder | Read the [architecture, state, HITL, recovery, and evaluation guide](docs/for-agent-harness-builders.md). |
  52. | Evaluating portability or trust boundaries | Review [compatibility](#7-compatibility), the [runtime contract](docs/runtime-contract.md), and [limitations](#8-limitations). |
  53. ---
  54. ## Contents
  55. 1. [Install — get the skills](#1-install--get-the-skills)
  56. 2. [Why this catalog exists](#2-why-this-catalog-exists)
  57. 3. [The pipeline — what each stage delivers to the next](#3-the-pipeline--what-each-stage-delivers-to-the-next)
  58. 4. [Use it](#4-use-it)
  59. 5. [See what each skill produces](#5-see-what-each-skill-produces)
  60. 6. [FAQ](#6-faq)
  61. 7. [Compatibility](#7-compatibility)
  62. 8. [Limitations](#8-limitations)
  63. ---
  64. ## 1. Install — get the skills
  65. The core asset is portable `SKILL.md`; Claude Code marketplace support is
  66. the convenience layer. Use the path that matches your host:
  67. | Host | Path |
  68. |---|---|
  69. | Claude Code | Marketplace install below; gets auto-triggering and `claude plugin list` verification. |
  70. | Codex CLI / Cursor / Windsurf | Load the canonical repo's `SKILL.md` into the host's skills/rules directory, or inline it into the prompt. |
  71. | Gemini CLI / generic API client | Inline `SKILL.md` as prompt or system context. |
  72. | Hermes | Install a raw `SKILL.md`; this repo has Hermes 0.13.0 skill-load verification for `literature-triage-matrix`. |
  73. | OpenClaw | Use a `SKILL.md` directory shape such as `~/.openclaw/skills/<skill>/SKILL.md` when supported; not release-verified here yet. |
  74. Full per-host recipes are in
  75. [docs/install.md → Using these skills outside Claude Code](docs/install.md#using-these-skills-outside-claude-code).
  76. **Claude Code fastest path — 30 seconds:**
  77. ```bash
  78. claude plugin marketplace add WenyuChiou/ai-research-skills
  79. claude plugin install research-workspace@ai-research-skills
  80. ```
  81. **Windows `cmd.exe`:** run the commands one at a time. If the plugin list
  82. does not show `research-workspace@ai-research-skills`, your paste probably
  83. only executed the first line; see [docs/install.md](docs/install.md) for the
  84. single-line CMD form.
  85. For Claude Code, the first two commands install the 12-skill
  86. `research-workspace` plugin. Add the optional plugins only when you need
  87. manuscript writing, full Zotero operations, or Claude-to-Codex/Antigravity
  88. delegation.
  89. **Additive install — stop after any step and use what you have:**
  90. ```bash
  91. # 1. Marketplace + 12 research-hub skills
  92. claude plugin marketplace add WenyuChiou/ai-research-skills
  93. claude plugin install research-workspace@ai-research-skills
  94. # 2. Manuscript work — outline, drafting, prose and evidence audit, scientific review (14/17)
  95. claude plugin install academic-writing-skills@ai-research-skills
  96. # 3. Zotero CRUD (15/17; enable local API in Zotero desktop first — docs/setup-guide.md §C)
  97. claude plugin install zotero-skills@ai-research-skills
  98. # 4. Multi-CLI delegation (17/17; install codex / agy CLI binaries first)
  99. claude plugin install codex-delegate@ai-research-skills
  100. claude plugin install antigravity-delegate@ai-research-skills
  101. # 5. Literature pipeline automation (Python CLI behind research-hub skills)
  102. pip install research-hub-pipeline
  103. research-hub setup --persona researcher
  104. research-hub doctor
  105. ```
  106. **Separate live smoke (not installation):**
  107. This is a **live** literature workflow, not an offline install check.
  108. `--no-nlm` disables NotebookLM only; it does not disable other network calls,
  109. models, or writes. Complete the [live-run preflight](docs/live-run-preflight.md)
  110. and obtain the applicable data/cost/write authorization before running:
  111. ```bash
  112. research-hub auto "agent-based modeling" --max-papers 3 --no-nlm
  113. ```
  114. Batch all 5 plugins in one go:
  115. ```bash
  116. bash scripts/install-all.sh # macOS / Linux / git-bash
  117. pwsh scripts/install-all.ps1 # Windows PowerShell
  118. ```
  119. **Verify:**
  120. ```bash
  121. claude plugin list
  122. # fastest path: research-workspace@ai-research-skills is ✔ enabled.
  123. # full install: 5 plugins ending in @ai-research-skills are ✔ enabled.
  124. ```
  125. `claude plugin list` verifies only the Claude Code marketplace install.
  126. It does not tell you whether Codex, Cursor, OpenClaw, Hermes, or a
  127. generic API client has loaded the `SKILL.md` files. Marketplace-installed
  128. plugins live under `~/.claude/plugins/cache/...`, not `~/.claude/skills/`.
  129. Per-plugin details: [docs/install.md](docs/install.md). Python / Zotero /
  130. Git not set up yet? Start with [docs/setup-guide.md](docs/setup-guide.md).
  131. Need to know whether a skill is prompt-only or needs the Python runtime?
  132. See [docs/runtime-contract.md](docs/runtime-contract.md).
  133. <details>
  134. <summary><b>I'd rather clone the repo</b> (contributors / debugging)</summary>
  135. ```bash
  136. git clone https://github.com/WenyuChiou/ai-research-skills.git
  137. cd ai-research-skills
  138. ```
  139. This catalog is the **registry**, not a monorepo. Each plugin's source
  140. code lives in its own repo:
  141. - `github.com/WenyuChiou/research-hub` — 12 `research-workspace` skills
  142. - `github.com/WenyuChiou/academic-writing-skills` — 2 skills
  143. - `github.com/WenyuChiou/zotero-skills` — 1 skill
  144. - `github.com/WenyuChiou/codex-delegate` — 1 skill
  145. - `github.com/WenyuChiou/antigravity-delegate` — 1 skill
  146. If you're hacking on a plugin, clone **its** source repo, not this
  147. catalog. This catalog only maintains `marketplace.json`, docs, image
  148. assets, and the catalog-side `CHANGELOG.md`.
  149. </details>
  150. ---
  151. ## 2. Why this catalog exists
  152. You probably know the AI-for-research pain points already. The five
  153. below are the ones this catalog actually fixes — not the ones it
  154. gestures at.
  155. ### P1 — "I keep re-explaining context to every new AI session"
  156. You open a new Claude / ChatGPT session to continue yesterday's work,
  157. and the model knows nothing about your research question, the
  158. experiments you've already finished, your baselines, or the gap you
  159. closed last week. You spend the first ten minutes re-typing all of it.
  160. Tomorrow, again.
  161. ### P2 — "AI confidently cites papers that don't exist"
  162. The model writes *"as Chen et al. (2024) demonstrated…"* — the paper
  163. is not real. You catch it (this time). In a reviewer response, a
  164. hallucinated citation is a desk-reject.
  165. ### P3 — "I've read 50 papers and still can't tell which gap is worth a thesis"
  166. Three questions need structured answers: *(1) is the gap actually
  167. still open? (2) is closing it a real contribution or a permutation of
  168. existing work? (3) can I close it in the time I have?* Tools that
  169. produce a one-paragraph "research gap summary" don't answer any of
  170. them.
  171. ### P4 — "AI-written prose smells like AI"
  172. *"Furthermore", "It is noteworthy that"*, hedged sentences with no
  173. committed claim. Reviewers (and senior co-authors) catch the smell in
  174. two paragraphs, and the manuscript drops in their priority list.
  175. ### P5 — "Switching Claude / Codex / Antigravity wipes my state"
  176. You design the prompt with Claude, hand it to Codex to scaffold the
  177. code, then use Antigravity only for bounded mechanical follow-up work.
  178. Long-context synthesis, CJK drafting, judgment, governance, and final
  179. review stay with the primary model. Each switch still costs five
  180. minutes of re-onboarding unless the handoff is explicit and durable.
  181. ---
  182. ### Three design principles, applied across 17 skills
  183. The catalog is arranged around three load-bearing ideas, not a feature
  184. list:
  185. | Principle | What it does | Solves |
  186. |---|---|---|
  187. | **1. Manifests** (`.research/`, `.paper/`) | Research state lives in checked-in YAML / Markdown files. A new AI session reads the manifest and re-onboards itself — you don't re-explain context. | P1, P5 |
  188. | **2. Schemas with anti-leakage rules** | Every cross-skill artifact has a YAML schema. A claim with empty `evidence_artifacts` is **forced** to carry `status: gap` + a `gap_reason` — never `supported`. A topic candidate with `verdict: do-not-pursue` is structurally separated from `worth-pursuing` ones. Downstream tools refuse to ship overconfident output. | P2, P3, P4 |
  189. | **3. Character-driven routing** | Token-heavy mechanical bulk → Codex. Bounded non-honesty-critical mechanical work → Antigravity. Long-context, CJK, judgment, governance, and final review stay with the primary model. The router (`research-hub-multi-ai`) writes a coordination file so each delegate reads its own brief, not the parent context. | P5 |
  190. The eight-stage visual map above shows these three principles applied
  191. to a real research workflow.
  192. ---
  193. ## 3. The pipeline — what each stage delivers to the next
  194. Eight stages from *"I should read about X"* to *"the manuscript
  195. shipped"*. Each stage's output is the next stage's input — the
  196. handoff is mechanical, not vibes.
  197. Use `research-workflow-orchestrator` when the whole lifecycle needs one
  198. stateful controller: it coordinates orient, scope, discover, synthesize,
  199. design, execute, write, and release; proceeds automatically for
  200. read-only or reversible work; and pauses at the five human gates.
  201. | # | Stage | Skill(s) | Output → next stage |
  202. |---|---|---|---|
  203. | 1 | **Discover literature** | `research-hub`, `paper-summarize` | `.bib` + per-paper Key Findings notes → Stage 2 |
  204. | 2 | **Find the gap** | `gap-to-topic`, `literature-triage-matrix`, `notebooklm-brief-verifier`, `zotero-library-curator` | `topic_dossier.gaps.yml` (with `verdict`, `verdict_reason`) → Stage 3a |
  205. | 3a | **Frame the RQ** | `research-design-helper` | `design_brief.md` (frontmatter `source: gaps.yml#G2`, `gap_verdict`) → Stage 3b + Stage 4 |
  206. | 3b | **Plan the project** | `research-context-compressor`, `research-project-orienter` | `project_manifest.yml` (`provenance.from_gap`) → Stages 4–8 |
  207. | 4 | **Build the model** | *cookbook* — `codex-delegate` for ≥5-file scaffold, Claude direct for ≤4-file or judgment work | code in your project repo (see [cookbook](docs/example-design-to-build.md)) → Stage 5 |
  208. | 5 | **Run & validate** | `research-context-compressor`, `research-project-orienter` | `.research/` run manifests so future AI sessions skip the rescan → Stage 6 |
  209. | 6 | **Visualise & interpret** | `codex-delegate`, `antigravity-delegate`, primary model | figures + analysis scripts → Stage 7 |
  210. | 7 | **Draft the manuscript** | `paper-memory-builder`, `academic-writing-skills` | `.paper/claims.yml` (with `status` enum + anti-leakage) → Stage 8 |
  211. | 8 | **Submit + respond** | `academic-writing-skills`, `research-context-compressor` | reviewer-response.md, version-tagged manifests → done |
  212. The cross-skill handoffs (Stage 2 → 3a → 3b → 8; Stage 7 → 8) are
  213. documented as **YAML schemas**, not free text. A downstream skill can
  214. refuse to process a malformed handoff — and does, when the schema
  215. violation would otherwise propagate (e.g. `status: gap` claims with no
  216. `gap_reason` are rejected at Stage 7).
  217. **Cross-cutting (every stage):** `research-workflow-orchestrator`,
  218. `codex-delegate`, `antigravity-delegate`, `research-hub-multi-ai`.
  219. These sit beside the pipeline, not on it — the orchestrator preserves
  220. lifecycle state and gates, while multi-agent routing uses Codex and
  221. Antigravity leaves for bounded mechanical execution.
  222. Full narrative + per-stage tool tables: [docs/pipeline.md](docs/pipeline.md).
  223. ---
  224. ## 4. Use it
  225. ### Pick by goal
  226. If you'd rather not read the full pipeline above, jump in by your
  227. immediate goal:
  228. | Your immediate goal | Skills you'll use |
  229. |---|---|
  230. | **Find & compare literature** | `research-hub` + `literature-triage-matrix` |
  231. | **Write or revise a paper** | `paper-memory-builder` + `academic-writing-skills` |
  232. | **Manage a research project** | `research-design-helper` + `research-context-compressor` + `zotero-library-curator` |
  233. For literature discovery, start with the search and source-reading capabilities
  234. actually available in Codex or Claude Code. Zotero, Obsidian and NotebookLM are
  235. optional, not a required setup detour; follow the existing
  236. [Stage 1 evidence handoff](docs/pipeline.md#1-discover-literature).
  237. > **Helping others adopt AI for research** (librarian / RA / advisor)?
  238. > No install needed — share this README plus [docs/install.md](docs/install.md).
  239. ### Or, just describe what you want
  240. Describe what you want in plain language — Claude Code matches your
  241. phrasing to a skill. You don't need to remember skill names.
  242. | When you say… | Skill that activates |
  243. |---|---|
  244. | "Compare these 5 papers by method, data, limitations" | `literature-triage-matrix` |
  245. | "Is this gap worth a thesis? Walk me through the three gates" | `gap-to-topic` |
  246. | "Walk me through my research design before I start coding" | `research-design-helper` |
  247. | "Check this paragraph's terminology, repeated wording, flow, and overclaim" | `academic-writing-skills` |
  248. | "Review this paper and load the relevant notation, display, SEM, LLM, water, or flood checks" | `paper-review` |
  249. Full trigger map (17 rows): [docs/skill-directory.md](docs/skill-directory.md).
  250. If auto-trigger picks the wrong skill, name it explicitly:
  251. *"Use `literature-triage-matrix` to compare these 5 papers."*
  252. ### Already use Zotero / Obsidian / NotebookLM?
  253. > **Have any of Zotero / Obsidian / NotebookLM in your workflow?**
  254. > [`research-hub`](https://github.com/WenyuChiou/research-hub) integrates
  255. > whichever of these you have as one library-management loop: search
  256. > new literature (arXiv, Semantic Scholar) → Zotero metadata + ingest
  257. > → per-paper notes synced to Obsidian → NotebookLM briefs verified
  258. > against source bundles. **All three tools are optional** —
  259. > `research-hub` adapts to whichever subset you have (or even none, in
  260. > sample-dashboard mode). Python-CLI-backed;
  261. > `pip install research-hub-pipeline`.
  262. >
  263. > Already run independent MCP servers for these tools (e.g.
  264. > `paper-search-mcp`, a Zotero MCP, a NotebookLM MCP)? They coexist —
  265. > the skill orchestrates workflows, the MCPs give raw tool access. The
  266. > CLI itself also exposes an MCP interface (`research-hub serve`).
  267. > Full setup + tool-by-tool modes:
  268. > [research-hub project README](https://github.com/WenyuChiou/research-hub).
  269. ### All 17 skills
  270. <details>
  271. <summary><b>From <a href="https://github.com/WenyuChiou/research-hub"><code>research-hub</code></a> (12 skills)</b> — one install gets all</summary>
  272. - [`research-hub`](https://github.com/WenyuChiou/research-hub/blob/master/skills/research-hub/SKILL.md) — search, ingest, organise papers across Zotero / Obsidian / NotebookLM. *(Stages 1, 2)*
  273. - [`literature-triage-matrix`](https://github.com/WenyuChiou/research-hub/blob/master/skills/literature-triage-matrix/SKILL.md) — comparison matrix across method, data, claim, limitation. *(Stage 2)*
  274. - [`notebooklm-brief-verifier`](https://github.com/WenyuChiou/research-hub/blob/master/skills/notebooklm-brief-verifier/SKILL.md) — verify NotebookLM briefs against source bundles. *(Stage 2)*
  275. - [`zotero-library-curator`](https://github.com/WenyuChiou/research-hub/blob/master/skills/zotero-library-curator/SKILL.md) — audit Zotero, propose cleanup (preview-only without `zotero-skills`). *(Stage 2)*
  276. - [`gap-to-topic`](https://github.com/WenyuChiou/research-hub/blob/master/skills/gap-to-topic/SKILL.md) — 3-gate go/no-go decision dossier for a candidate thesis/proposal topic (open? a contribution? feasible?). Emits `.gaps.yml` for `research-design-helper` Stage 3a handoff. *(Stage 2)*
  277. - [`research-design-helper`](https://github.com/WenyuChiou/research-hub/blob/master/skills/research-design-helper/SKILL.md) — Socratic dialog through RQ → mechanism → identifiability → validation → risk. Reads `.gaps.yml` to pre-fill segments 1 + 5; Stage 4 [cookbook](docs/example-design-to-build.md) reuses the produced `design_brief.md`. *(Stages 3a, 4)*
  278. - [`research-context-compressor`](https://github.com/WenyuChiou/research-hub/blob/master/skills/research-context-compressor/SKILL.md) — `.research/` manifests so future AI sessions skip the rescan. *(Stages 3b, 5, 8)*
  279. - [`research-project-orienter`](https://github.com/WenyuChiou/research-hub/blob/master/skills/research-project-orienter/SKILL.md) — fast orientation memo from those manifests. *(Stages 3b, 5)*
  280. - [`research-hub-multi-ai`](https://github.com/WenyuChiou/research-hub/blob/master/skills/research-hub-multi-ai/SKILL.md) — character-driven routing across the primary model, Codex, and Antigravity. *(Cross-cutting)*
  281. - [`research-workflow-orchestrator`](https://github.com/WenyuChiou/research-hub/blob/master/skills/research-workflow-orchestrator/SKILL.md) — resumable controller for orient → scope → discover → synthesize → design → execute → write → release, automatic for read-only/reversible work and paused at catalog human gates. *(All stages)*
  282. - [`paper-memory-builder`](https://github.com/WenyuChiou/research-hub/blob/master/skills/paper-memory-builder/SKILL.md) — `.paper/claims.yml` + `.paper/figures.yml` (status enum, anti-leakage rule, file sentinels). *(Stage 7)*
  283. - [`paper-summarize`](https://github.com/WenyuChiou/research-hub/blob/master/skills/paper-summarize/SKILL.md) — fill per-paper Key Findings / Methodology / Relevance in both Obsidian and Zotero child notes after `research-hub auto`. *(Stage 1)*
  284. </details>
  285. <details>
  286. <summary><b>Standalone repos (4 plugins)</b> — one plugin install each</summary>
  287. - [`academic-writing-skills`](https://github.com/WenyuChiou/academic-writing-skills/blob/main/skills/academic-writing-skills/SKILL.md) — extended outlines, scoped pre-draft plans and post-edit checks, chapter closure, shared terminology records, four-pass manuscript review, cross-artifact revision, and release readiness. *(Stages 7, 8)*
  288. - [`paper-review`](https://github.com/WenyuChiou/academic-writing-skills/blob/main/skills/paper-review/SKILL.md) — general scientific review with progressively selected equation/display provenance, psychometrics/SEM, AI/LLM, water/CNHS, flood/hydrodynamics, round, and explicit reviewer modules. *(Stages 7, 8)*
  289. - [`zotero-skills`](https://github.com/WenyuChiou/zotero-skills/blob/master/skills/zotero-skills/SKILL.md) — full Zotero CRUD, batch metadata, library maintenance. *(Stages 1, 2, 7)*
  290. - [`codex-delegate`](https://github.com/WenyuChiou/codex-delegate/blob/master/skills/codex-delegate/SKILL.md) — Claude → Codex CLI handoff for code-heavy / mechanical work. *(Cross-cutting, also Stages 4, 6)*
  291. - [`antigravity-delegate`](https://github.com/WenyuChiou/antigravity-delegate/blob/master/skills/antigravity-delegate/SKILL.md) — Claude → Antigravity CLI handoff for bounded non-honesty-critical mechanical work. *(Cross-cutting, also Stages 4, 6)*
  292. </details>
  293. ### Time + cost expectations
  294. Rough envelope from in-session use — adjust to your input size:
  295. | Task | Typical wall time | Conversation turns | Notes |
  296. |---|---|---|---|
  297. | Compare 5 papers (`literature-triage-matrix`) | 1–3 min | 1–2 | Linear in paper count; 20 papers ≈ 5 min |
  298. | 3-gate gap decision (`gap-to-topic`) | 5–15 min | 3–6 | Scales with candidate count + literature-recall depth |
  299. | Terminology, repetition, and flow audit on 1 paragraph (`academic-writing-skills`) | <1 min | 1 | Independent of manuscript size |
  300. | Reviewer response (6 comments) (`academic-writing-skills`) | 3–8 min | 3–5 | Scales with comment depth + revision required |
  301. | Audit 800-item Zotero library (`zotero-library-curator`) | 2–4 min | 1 | Read-only; library size matters less than tag diversity |
  302. | Summarize 5 papers per cluster (`paper-summarize`) | 4–10 min | 1 | One LLM call per paper; rolls back per-paper on failure |
  303. These are **maintainer-observed ranges**, not benchmark numbers. Your
  304. LLM provider, network, library state, and prompt phrasing all affect
  305. the actual time.
  306. ### ⚠ Back up Zotero before any CRUD
  307. `zotero-library-curator` is read-only — it emits a preview report.
  308. `zotero-skills` *can* write (merge duplicates, delete items, rebind
  309. collections). **Always export a Zotero backup before letting any AI
  310. modify your library**: Zotero → File → "Export Library…" → Zotero RDF.
  311. The skills will not do this for you.
  312. ---
  313. ## 5. See what each skill produces
  314. Each pipeline stage's deliverable has a worked-example file. Click
  315. through to see the actual artifact a skill emits — not a description.
  316. | Stage | What the skill produces | Example file |
  317. |---|---|---|
  318. | 1–2 | Literature-review deliverable (TL;DR + per-paper synthesis + tagged gap analysis) | [`example-literature-review-deliverable.md`](docs/example-literature-review-deliverable.md) · [`.docx`](docs/example-literature-review-deliverable.docx) |
  319. | 2 | Topic dossier with the 3-gate decision (open / contribution / feasibility) + machine-readable gap roster | [`example-topic-dossier.md`](docs/example-topic-dossier.md) + [`.gaps.yml`](docs/example-topic-dossier.gaps.yml) · [`.docx`](docs/example-topic-dossier.docx) |
  320. | 3a | Design brief with provenance to the dossier, Socratic-walked RQ → mechanism → identifiability → validation → risk | [`example-design-brief.md`](docs/example-design-brief.md) |
  321. | 3b | Project manifest carrying `provenance.from_gap` so future AI sessions skip the rescan | [`example-project-manifest.yml`](docs/example-project-manifest.yml) |
  322. | 4 | Cookbook — two paths from brief to scaffolded code (Claude-direct for ≤4 files; `codex-delegate` for ≥5) | [`example-design-to-build.md`](docs/example-design-to-build.md) |
  323. | 7 | Paper claims with `status: gap` anti-leakage rule + figure roster with embedded-in-manuscript sentinels | [`example-paper-memory-claims.yml`](docs/example-paper-memory-claims.yml) + [`example-paper-memory-figures.yml`](docs/example-paper-memory-figures.yml) |
  324. The cross-cutting and orchestration skills (`research-hub` ingestion,
  325. `research-project-orienter`, `research-hub-multi-ai`,
  326. `notebooklm-brief-verifier`) don't map to a single stage deliverable —
  327. they're exercised end-to-end in [`test-corpus/`](test-corpus/) from a
  328. real dogfood run: an
  329. [orientation memo](test-corpus/ai-agents-social-interaction/.research/orientation_memo.md),
  330. a [multi-AI routing decision](test-corpus/ai-agents-social-interaction/.research/multi-ai-routing-decision.md),
  331. and a [NotebookLM brief verification](test-corpus/ai-agents-social-interaction/.research_hub/brief-verify-20260426T001559Z.md).
  332. Or read [`docs/examples.md`](docs/examples.md) end to end for the
  333. narrative tour with per-skill input/output tables.
  334. ---
  335. ## 6. FAQ
  336. ### Does this work with Chinese-language literature?
  337. Yes — skills are language-agnostic. Run them in Chinese, the output is
  338. Chinese. NotebookLM, Claude, and Gemini all handle Traditional /
  339. Simplified Chinese well.
  340. ### How do I export to `.docx` for an advisor?
  341. Stage 2 dossier and Stage 1–2 literature-review deliverable ship with
  342. `.docx` generators — direct links in §5 above. Other stages
  343. (`design_brief.md`, `project_manifest.yml`, `claims.yml`) emit
  344. Markdown / YAML; pipe through `pandoc` or Word's "Open Markdown" if
  345. needed.
  346. ---
  347. ## 7. Compatibility
  348. The portable layer is the `SKILL.md` instruction file plus any bundled
  349. `references/`, `scripts/`, and workflow contracts. Claude Code
  350. marketplace support is one host-specific packaging layer, not the only
  351. way to use the skills.
  352. | Layer | What is portable | Status |
  353. |---|---|---|
  354. | Universal `SKILL.md` layer | Skill instructions, trigger descriptions, references, scripts, and project handoff contracts | 17 catalogued skills; recorded source-layout/manifest checks do not establish live host behavior |
  355. | Host-specific behavior | Auto-triggering, plugin marketplace install, `claude plugin list`, skill discovery, and rules-directory conventions | Depends on the agent host; use that host's own list/discovery check |
  356. | Current portability audit | Generic `SKILL.md`-loading hosts | 11/14 zero-edit portable in the 2026-05-10 audit; 3/14 needed cosmetic `<skill-root>` path edits that have since landed |
  357. | Verified host install | NousResearch/hermes-agent 0.13.0 | `literature-triage-matrix` installed end to end, security scan SAFE, registered `enabled`; Hermes inference loop not tested |
  358. | OpenClaw | `SKILL.md`-style directories such as `~/.openclaw/skills/<skill>/SKILL.md` when supported by the user's OpenClaw install | Structurally compatible target, but not release-grade verified by this repo yet |
  359. | Other agents | Codex CLI, Gemini CLI, Cursor, Windsurf, generic API clients, and other `SKILL.md`-loading hosts | Load the same `SKILL.md` as context or into the host's skill/rules directory; not all hosts are individually tested |
  360. The `11/14` portability figure reflects the audit run on 2026-05-10,
  361. when the catalog had 14 skills. Later additions (`gap-to-topic`,
  362. `paper-review`, `research-workflow-orchestrator`, and `antigravity-delegate`)
  363. have separate catalog records and were not part of that portability audit.
  364. Calibrated audit + experiment transcripts:
  365. [`.research/hermes-compatibility-audit.md`](.research/hermes-compatibility-audit.md).
  366. For documented Codex CLI / Gemini CLI / Cursor / Windsurf / Hermes and
  367. generic-API examples, see
  368. [docs/install.md → Using these skills outside Claude Code](docs/install.md#using-these-skills-outside-claude-code).
  369. For OpenClaw, use the `SKILL.md` directory shape above until this repo adds
  370. release-grade OpenClaw verification.
  371. ---
  372. ## 8. Limitations
  373. - Assembled and tested by one graduate-student researcher; not
  374. corpus-scale-validated.
  375. - Domain bias toward water resources and agent-based modeling; not
  376. validated for social sciences, ML, or clinical writing.
  377. - Behavioral correctness on real-world inputs is the source repo's
  378. responsibility, not this catalog's.
  379. - Report-only monthly health/drift checks inspect upstream repositories,
  380. manifests, skill URLs, and verification-age records. A successful HTTP
  381. response or source-layout check does not establish live host behavior or
  382. renew a human verification date; unresolved findings need human review.
  383. - No `claude plugin install` round-trip is asserted by CI; the
  384. marketplace registry is checked structurally, the actual install +
  385. trigger path is verified by the maintainer between releases (see
  386. [docs/verification.md](docs/verification.md) for what is and isn't
  387. covered).
  388. - Current research-workspace source no longer embeds `zotero-skills`; the
  389. standalone plugin is canonical. The bare-name collision in
  390. [the 2026-05-20 verification](docs/verification.md#2026-05-20--phase-53b-end-to-end-verification)
  391. is historical. Old host caches can differ from current source, so inspect
  392. your installed plugin state before treating the source fix as a local update.
  393. The full design contract — including what is and is not machine-checked
  394. — is in [docs/design-philosophy.md](docs/design-philosophy.md).
  395. ---
  396. ## License
  397. MIT. Each skill is maintained in its canonical repo — this catalog is
  398. the index, not a monorepo. Contributions welcome via issue or PR.
  399. New-skill proposals → either `research-hub` (workflow integration) or
  400. a standalone repo (deep, single-purpose CRUD).

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.