research-project-router

Research Project Router

Use this skill first when a request spans more than one research workflow. The goal is to route deliberately and preserve project state.

Read First

• references/repository-contract.md
• references/output-contracts.md

Routing Table

User intent	Prefer skill
Turn ideas, gaps, hunches, and vague goals into research questions, contribution claims, scope, and evidence plans	research-design-positioning
Choose CS venue, track, audience, review criteria, or paper positioning	cs-venue-strategy
Add PDFs, papers, proposals, archives, datasets, or web pages	source-ingestion
Convert PDFs, scans, DOCX, tables, or figures to Markdown/assets	document-conversion
Normalize BibTeX, identifiers, citation keys, missing/unused citations	citation-bibliography-tooling
Find related work, build SOTA, compare papers, discover gaps	sota-literature-review
Build a SOTA from an idea, seed, or few references into full-text syntheses, BibTeX, survey LaTeX, and research agenda	research-design-positioning then academic-mcp-tooling then sota-literature-review then citation-bibliography-tooling then citation-claim-audit
Run reproducible systematic/scoping review and PRISMA-style screening	systematic-review-prisma
Turn reviewed SOTA claims into survey sections and survey claim ledgers	survey-synthesis
Prioritize SOTA gaps and survey claims into research opportunities	research-agenda
Create contribution packages, claim maps, reports, badge plans, and paper exports	contribution-package
Write reports from completed analyses and generated outputs	research-results-reporting
Prepare publication figures, tables, captions, source data, and asset maps	publication-figures-tables
Check citations, claims, numbers, evidence, bibliography	citation-claim-audit
Design/audit CS evaluation, baselines, metrics, ablations, validity	cs-methodology-evaluation
Select an accepted paper frame from contributions, evidence, venue fit, and badge fit	paper-framing
Build paper-specific release manifests, source maps, locks, checksums, and staged artifacts	paper-release
Review a manuscript/proposal as a severe peer reviewer	adversarial-peer-review
Draft, revise, polish, or review a paper/proposal	paper-writing-review
Manage cover letters, submission checklists, submitted locks, decisions, and camera-ready state	paper-submission-lifecycle
Respond to reviews, rebuttal, revision plan, camera-ready changes	rebuttal-revision-strategy
Audit badge, open-science, release, transparency, and venue compliance evidence	badge-compliance-profiles
Prepare artifact evaluation, code/data release, reproducibility package	artifact-open-science
Migrate a bad repo, archive, or notebook pile into the project structure	repo-migration
Understand or reproduce an external research repo	research-repo-reproduction
Run experiments, benchmark ideas, maintain run ledger	experiment-logbook
Analyze datasets, statistics, tables, and figures	research-data-analysis
Build research dashboards or interactive demos	research-ui-prototyping
Assess human-subject, privacy, scraping, license, or sharing risk	ethics-data-governance
Fix folder structure, docs, wiki, agent instructions	research-project-maintenance
Onboard or audit the skill/MCP capability stack	academic-mcp-tooling then research-project-maintenance
Select, configure, smoke-test, and document arXiv, DBLP, PubMed, Semantic Scholar, OpenAlex, Zotero, Overleaf, Crossref, Scholar, and scholarly metadata MCP tools	academic-mcp-tooling
Review or improve the skill package itself	skill-evaluation

Default Procedure

1. Identify the research artifact being touched: source, claim, data, code, experiment, paper, or project structure.
2. Check whether the repository follows the contract. If not, adapt conservatively and note the gap.
3. Route to the narrowest skill that can complete the task.
4. If several workflows are required, order them as: source ingestion -> SOTA -> survey -> research agenda -> contribution -> analysis -> paper framing -> paper release + manuscript -> submission -> response/revision.
5. Update wiki/log.md or the appropriate output ledger when the task changes project knowledge.

Maintain the Project Quality Contract throughout routing: classify the request, keep outputs in the correct work zone, promote only trusted outputs, and update the relevant ledger, wiki page, or badge evidence record when the task creates durable project state.

For full SOTA or survey work in a create-academic-research project, run npm run workflow:literature early so arXiv, DBLP, Semantic Scholar, and OpenAlex graph search are selected before citation chasing.

Do Not

• Treat chat history as the project record.
• Skip ingestion and write a SOTA directly from raw PDFs.
• Let exploratory experiment outputs overwrite trusted reproduction outputs.
• Make a claim stronger than the source, dataset, or run supports.