Install
$ agentstack add skill-neuroanalytics-data-science-harness-preregister ✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README — it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming — see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps — measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
Skill: preregister
Turn a comparison into a confirmatory one: freeze its specification before any data is analyzed, register that spec with an external registry, and record it as an outstanding obligation in the ledger. The frozen spec is the STAMPED spec-centric research object (Self-contained, Tracked, Actionable) — the durable thing; the later run is Ephemeral and is checked against this spec. You own the freeze/registration judgment; you delegate the save to the datalad doer.
> This is the confirmatory path of analyze/propose-comparison. A "preregistration" recorded after > looking at the outcome is not one — the freeze must precede execution. analyze/run-comparison > then checks results against the registered spec and any deviation is reportable.
When to use
- A comparison should be confirmatory: its hypothesis, inputs, and analysis plan are to be frozen
and registered before running (often right after propose-comparison picks the confirmatory mode).
- Do NOT use for exploratory quick queries (just
propose-comparison), or to execute the analysis
(run-comparison), or to resolve/track existing obligations (govern/obligations).
Steps
- Freeze the spec — capture the immutable specification for the comparison's
cmp/
branch: what (hypothesis), why, inputs, expected_outputs, and the analysis plan (model, primary outcome, exclusions, corrections). Write it as a durable, versioned file (e.g. code/prereg/.md) so the frozen text is itself tracked. Confirm with the user that this is final — after registration it does not change.
- Register externally — guide the user to submit the frozen spec to the appropriate registry
(OSF Registrations, ClinicalTrials.gov, PROSPERO for systematic reviews) and obtain the registration id/URL. The harness records the reference the user provides; it does not fabricate or auto-submit one. If registration is pending, record the obligation as pending with no ref yet and note that the ref must be added once assigned.
- Record the obligation — add to
project.yamlobligations[](perdocs/project-ledger.md):
{ id: prereg-, kind: preregistration, description: "; frozen for cmp/", due: , status: pending, ref: }.
- Log it — append
{ ts, op: preregister, stage: govern, note: "froze + registered cmp/ ()", branch: cmp/ }. - Save — delegate to the datalad doer: "save:
datalad save -m 'preregister cmp/: freeze spec + register ()'." - Report — the frozen spec path, the registration ref (or that it is pending), the new pending
obligation, and that run-comparison must check outcomes against this spec — deviations are reportable. The obligation is discharged (via govern/obligations) once the confirmatory comparison is completed as specified.
Constraints
- Freeze precedes execution — never "preregister" a comparison whose outcome has already been seen;
if it has, record it honestly as exploratory (propose-comparison), not confirmatory.
- The frozen spec is immutable after registration — a change means a new comparison + a new
registration, recorded as such.
- Never fabricate a registration id/URL — record only what the registry actually assigned.
- Add the obligation per the ledger conventions; keep
log:append-only and the ledger
schema-valid. Delegate the save to the datalad doer.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: neuroanalytics
- Source: neuroanalytics/data-science-harness
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet — be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.