agentleFS
leiverkusleiverkus/research-superpowers/skills/finishing-a-research-project/SKILL.md

finishing-a-research-project

Skill4 starsChanged 5 months agoMITView on GitHub

Use when a research project reaches completion — manuscript finalized, peer-review cleared, ready for publication / submission / archive. Runs the final checklist across rendering, BibTeX, wiki, data, reproducibility, archive, and follow-up grant planning.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 2 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Before declaring "done", check every item on the Closing Checklist below.

    line 13
  2. Very specific:

    If done done: commit everything (git add .

    line 39
  3. Very specific:

    Check author guidelines — may require format conversion (quarto render --to docx / --to latex, or pandoc)

    line 114
  4. Distinctive:

    Do NOT submit on the user's behalf — prepare and hand over

    line 117

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Finishing a Research Project

The project is done only when the last checklist is green. This skill walks the final mile: rendering, archiving, DOI, submission, and optional grant follow-up.

Announce at start: "Using finishing-a-research-project to close out <slug>."

<SOFT-GATE> Before declaring "done", check every item on the Closing Checklist below.

For unmet items: tell the user which, ask for a short reason per skipped item, write reasons into knowledge/_meta/gate-overrides.log, and mark the project in the log as either status: done (with documented omissions) or status: paused. Hidden partial completion is not closure — explicit justification is. </SOFT-GATE>

Closing Checklist

  1. Manuscript renders cleanly — make render in output/<book|article>/ exits 0; PDF/HTML opens without warnings
  2. Wiki lint green — python scripts/lint-wiki.py exits 0
  3. BibTeX consistent — every citation key in the manuscript exists in references.bib; no unused keys (or unused are intentional)
  4. Peer review done — both constructive and adversarial reviewer reports archived under output/<target>/reviews/; all Major issues resolved or explicitly deferred with rationale
  5. Hypothesis addressed — pre-registered Hypothesis is explicitly confirmed, refuted, or marked "inconclusive — exploratory" in the manuscript
  6. Data reproducibility — output/data-analysis/ has requirements.txt or environment.yml; a README says "run python main.py to reproduce"; seeds are fixed
  7. Data archive — raw datasets in input/data/ are licensed; derivatives in output/data-analysis/results/ are documented
  8. Synthesis stable — all knowledge/synthesis/*.md referenced by the manuscript have status: stable
  9. Log closing entry in knowledge/_meta/log.md:
    - YYYY-MM-DD · finish · <slug> · output=<path>
  10. Publication archive (DOI) — offer to deposit on Zenodo (or institution repository); update the log with DOI
  11. Decision: submission / follow-up — journal / publisher / conference? grant renewal?
  12. If follow-up grant: route to grant-finder
  13. If done done: commit everything (git add . && git commit -m "finish: <slug>"), tag the release

Process Flow

digraph finishing {
    "Render check" [shape=box];
    "Wiki lint" [shape=box];
    "BibTeX consistency" [shape=box];
    "Peer review archived" [shape=box];
    "Hypothesis resolution" [shape=box];
    "Data reproducibility" [shape=box];
    "Data archive" [shape=box];
    "Synthesis stable" [shape=box];
    "Any check fail?" [shape=diamond];
    "Fix and re-check" [shape=box];
    "Log closing entry" [shape=box];
    "DOI / archive publication" [shape=box];
    "Follow-up grant?" [shape=diamond];
    "Invoke grant-finder" [shape=box];
    "Commit & tag" [shape=box];
    "Done" [shape=doublecircle];

    "Render check" -> "Wiki lint";
    "Wiki lint" -> "BibTeX consistency";
    "BibTeX consistency" -> "Peer review archived";
    "Peer review archived" -> "Hypothesis resolution";
    "Hypothesis resolution" -> "Data reproducibility";
    "Data reproducibility" -> "Data archive";
    "Data archive" -> "Synthesis stable";
    "Synthesis stable" -> "Any check fail?";
    "Any check fail?" -> "Fix and re-check" [label="yes"];
    "Fix and re-check" -> "Render check";
    "Any check fail?" -> "Log closing entry" [label="no"];
    "Log closing entry" -> "DOI / archive publication";
    "DOI / archive publication" -> "Follow-up grant?";
    "Follow-up grant?" -> "Invoke grant-finder" [label="yes"];
    "Follow-up grant?" -> "Commit & tag" [label="no"];
    "Invoke grant-finder" -> "Commit & tag";
    "Commit & tag" -> "Done";
}

Reproducibility Statement Template

Drop into the manuscript (or supplementary material):

## Reproducibility

All analyses are reproducible from `output/data-analysis/`:

- **Code:** <files>
- **Data inputs:** `input/data/<files>` (licenses: <list>)
- **Environment:** Python <version>, see `output/data-analysis/requirements.txt`
- **Entry point:** `python output/data-analysis/main.py`
- **Random seed:** <fixed value>
- **Repository:** <git URL or Zenodo DOI>

Archive / DOI

Options:

  • Zenodo — free, accepts datasets + code + manuscripts; gives DOI immediately
  • Institutional repository — university OA repository
  • GitHub Release + Zenodo-GitHub integration — tag release, Zenodo creates DOI automatically

Pick one with user, update log entry with DOI.

Submission Handoff

If submission targeted:

  • Identify journal / publisher / conference CfP
  • Check author guidelines — may require format conversion (quarto render --to docx / --to latex, or pandoc)
  • Produce cover letter draft
  • Confirm corresponding author + ORCID
  • Do NOT submit on the user's behalf — prepare and hand over

Red Flags

ThoughtReality
"Wiki-lint is just a detail, I'll skip it"Broken wikilinks are the #1 reviewer feedback point.
"The hypothesis was wrong, I'll just reword it"Pre-registration principle: document honestly. The result is what it is.
"I'll do the Zenodo DOI later"Later = never. Now, or explicitly into the log.
"Follow-up grants are separate projects"Now is the best moment — the story is fresh in your head.
"It's just a working paper, not a publication"Even unpublished work must be reproducibly archived.

Key Principles

  • The checklist is the closure — not "feels done"
  • Hypothesis honesty — what was predicted, what came out, what's open
  • Reproducibility — manuscript AND data AND code AND environment
  • DOI before submission — an archived version exists before the publisher process
  • Log entry is the final artefact — the project is closed in the log

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.