agentleFS
fabian-von-tiedemannfabian-von-tiedemann/research-harness-template/.agents/skills/new-investigation/SKILL.md

new-investigation

Skill1 starsChanged 2 months agoMITView on GitHub

Start a new investigation in this research repo: directory, README, first study protocol, register check. Use when the researcher has a question to test. Not for editing an existing investigation.

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 1 is distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read method/README.md, method/templates.md and investigations/README.md.

    line 12
  2. Distinctive:

    Ask one thing at a time, and write each answer down before the next question:

    line 14
  3. Very specific:

    Create investigations/NNN-<slug>/README.md from the investigation template with those answers.

    line 21
  4. Very specific:

    Create investigations/NNN-<slug>/study-01/protocol.json with id: INNN-01, investigation: INNN, the measures and the rule.

    line 22
  5. Very specific:

    Create study-01/README.md from the study template, status protocol written.

    line 23
  6. Very specific:

    Add a row to investigations/README.md.

    line 24

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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

New investigation

You create the skeleton for one investigation with one study, following method/templates.md. You do not gather data or write conclusions.

Steps

  1. Read method/README.md, method/templates.md and investigations/README.md.
  2. Find the next number: the highest NNN in investigations/ plus one. Ask for a slug if the question does not give an obvious one.
  3. Ask one thing at a time, and write each answer down before the next question:
    • The question, as one sentence.
    • The hypothesis: what would be true if the answer is yes.
    • The competing hypothesis: what would be true if the answer is no, in a form data could support. If the researcher says "the hypothesis is false", push once for something a study could distinguish.
    • The first study: what it measures, in what unit, and the two thresholds. They must not overlap. If the researcher has no thresholds because the study is exploratory, set exploratory: true and interpretation_rule: null, and say in the study README why.
    • Data kind: simulated, primary or secondary.
    • Declared interests: funding, CRediT roles, conflicts of interest, prior position. "None" only if true.
  4. Create investigations/NNN-<slug>/README.md from the investigation template with those answers. Status planned.
  5. Create investigations/NNN-<slug>/study-01/protocol.json with id: INNN-01, investigation: INNN, the measures and the rule. inputs lists the files the study will read under study-01/inputs/; create the directory with a .gitkeep if no inputs exist yet. affects is empty until a claim exists.
  6. Create study-01/README.md from the study template, status protocol written.
  7. Add a row to investigations/README.md.
  8. Run python3 -m harness validate.
  9. Tell the researcher what was created, and that the next steps are: put the input files in study-01/inputs/, run python3 -m harness create investigations/NNN-<slug>/study-01/protocol.json --out investigations/NNN-<slug>/study-01/frozen, and commit frozen/ before any analysis. That commit is the pre-registration record.

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.