By fdhidalgo
Agent skills for social-science research workflows: setup, Socratic grilling against project docs, analysis-plan synthesis, and decomposition into a DAG of GitHub issues. R + targets first; Python + marimo later.
One-time per-repo setup for the research-skills suite. Writes an `## Agent skills` block to `CLAUDE.md`, seeds `docs/agents/*.md`, `CONTEXT.md`, and `docs/adr/0000-template.md`, and creates the GitHub labels (four triage-state labels + the `epic` and `autonomous-ok` kind labels). Run before first use of `grill-with-docs-research`, `to-analysis-plan`, or `to-issues-research` — or if those skills appear to be missing context about this repo's issue tracker, triage labels, or research docs.
Synthesise the current conversation, `CONTEXT.md`, and ADRs into a numbered analysis-plan document at `docs/plans/NNNN-slug.md`. Versioned-snapshot model — every invocation writes a new file; successors reference the prior plan and what changed. Does NOT interview the user; if context is thin, surfaces gaps as a first-class "Open questions" section and points back to `grill-with-docs-research`. Use when the user wants to lock a plan after grilling, draft a successor plan after a referee report / data refresh / identification pivot, or get a snapshot of the current state of thinking.
Decompose an analysis plan at `docs/plans/NNNN-slug.md` into a DAG of GitHub Issues using the `Depends on: #N` convention from `docs/agents/issue-tracker.md`, sized so each issue is plausibly a single session. Publishes in dependency order, applies state labels and the `autonomous-ok` kind label where appropriate, and populates the matching epic's `Children:` placeholder (or offers once to create an epic). Surfaces the plan's Open questions item-by-item and converts the ones the user opts in to `needs-info` issues. Use when the user wants to turn a locked analysis plan into pickable work, populate an epic with child-issue references, or convert specific open questions into actionable tickets.
Socratic grilling session that stress-tests a research plan against the project's recorded language (`CONTEXT.md` — glossary, variables, population) and methodology ADRs (`docs/adr/*.md`), updating both inline as decisions crystallise. Use when the user wants to grill a research plan, hypothesis, estimator choice, sample restriction, identification strategy, prior, or other methodological commitment.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
A Claude Code plugin: a suite of agent skills for social-science research workflows (R + targets primary, Python + marimo later).
Adapted from Matt Pocock's skills and re-targeted at research artifacts (analysis plans, ADRs, data validation, probes) instead of software-product artifacts (PRDs, features, tests).
The suite serves two project shapes equally: papers built around a research question, and standalone dataset-creation projects that scrape and combine sources into a released dataset.
Design complete, implementation in progress. See PLAN.md for the full design and build sequence.
Shipped so far:
setup-research-skills — one-time per-repo setup. Writes the ## Agent skills block to CLAUDE.md, seeds docs/agents/*.md, CONTEXT.md, and docs/adr/0000-template.md, and creates the GitHub labels (four state + the epic and autonomous-ok kind labels).grill-with-docs-research — Socratic grilling session that stress-tests a research plan against the project's recorded language (CONTEXT.md glossary / variables / population) and methodology ADRs. Writes inline updates as terms and decisions crystallise.to-analysis-plan — synthesises the current conversation + CONTEXT.md + ADRs into a numbered analysis-plan file at docs/plans/NNNN-slug.md. Versioned snapshots (successors cite predecessors); does NOT interview; surfaces gaps as a first-class "Open questions" section. Offers once to also create a top-level GitHub epic issue.to-issues-research — decomposes the latest analysis plan into a DAG of GitHub Issues using Depends on:, sized so each issue is plausibly one session of work. Applies state labels and the autonomous-ok kind label per task category, walks the plan's Open questions item-by-item, and populates (or offers to create) the matching epic's Children: list.Coming next (in build order — see PLAN.md §6): run the thin slice on a real research project before deepening any single skill. The full 14-skill catalog is in PLAN.md §4.
In any Claude Code session:
/plugin marketplace add fdhidalgo/research_skills
/plugin install research-skills@research-skills
This adds the repo as a marketplace, then installs the four skills above. Run /plugin marketplace update research-skills to pull the latest.
0BSD — public-domain-equivalent, no attribution required.
npx claudepluginhub fdhidalgo/research_skills --plugin research-skillsUpstash Context7 MCP server for up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your LLM context.
A growing collection of Claude-compatible academic workflow bundles. Covers scientific figures, manuscript writing and polishing, reviewer assessment, citation retrieval, data availability, paper reading, literature search, response letters, paper-to-PPTX conversion, and evidence-grounded Chinese invention patent drafting. Rules are organized as reusable skill folders with explicit workflows and quality checks.
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
Harness-native ECC operator layer - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses
Plugin-safe Claude Code distribution of Antigravity Awesome Skills with 1,561 supported skills.
Upstash Context7 MCP server for up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your LLM context.