Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/whenpoem/aiscientist/preregisternpx skills add whenpoem/aiscientist --skill preregistergit clone --depth 1 https://github.com/whenpoem/aiscientistWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/whenpoem/aiscientist/preregister)<a href="https://agentmods.dev/skills/whenpoem/aiscientist/preregister"><img src="https://agentmods.dev/badge/skills/whenpoem/aiscientist/preregister.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00062 | $0.00866 |
| Opus 5 | $0.00031 | $0.00433 |
| Sonnet 5 | $0.00012 | $0.00173 |
| Haiku 4.5 | $0.00006 | $0.00087 |
Grade A, and why
preregister scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 58 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Preregister
This skill supports "decide before observing" for confirmatory claims. Exploratory runs may still use seed_perturb / pin_metric, but any number intended as a main publication claim should have a matching prereg row first.
When to invoke
- Before a result is promoted from exploratory to a confirmatory manuscript claim.
- Right after
bt-tournamentreturns the top-2 hypotheses when the next run is explicitly confirmatory. - Whenever the user types
/preregister.
Required arguments
| arg | meaning | example |
|---|---|---|
hypothesis_id |
id from mem_nodes (kind=hypothesis) |
hyp_a3f9... |
metric_name |
exact claim text the engineer will pin later | "test accuracy" |
direction |
higher_better or lower_better |
higher_better |
threshold |
number that separates met from missed |
0.85 |
seed_count |
how many seeds the seed_perturb call must use | 5 |
alpha |
nominal Type-I error rate | 0.05 |
mc_correction |
bonferroni (default), none, or legacy alias bh |
bonferroni |
family_id |
stable id shared by related confirmatory tests | primary_metrics |
family_size |
total number of tests planned in that family | 4 |
heldout_dataset |
optional dataset name to be queried via query_heldout | mnist-test |
Workflow
- Call
mcp__verify__list_preregistrations(hypothesis_id=...). If a row exists with statusopen, do not fail the session. Ask whether to reuse that lock, withdraw it outside this tool, or create a separate confirmatory prereg for a genuinely different metric. - Before the first family member is resolved, define the full family. Use one
family_idand the same lockedfamily_size, alpha, and correction for all related tests. A standalone test gets an automatically generated family of 1. - Call
mcp__verify__preregister(...)with the exact metric text, threshold,family_id, andfamily_size. - Surface the resulting
prereg_idand family metadata before implementation. - Pass
prereg_idalong the workflow so the confirmatory run stays linked.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 4d ago First seen · 58 lines · 62 tokens per session scan A afb1e3b30d1d
preregister is a skill published in the GitHub repository whenpoem/aiscientist (8 stars, last pushed 1mo ago), licensed MIT. It adds 62 tokens to every session and 866 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
generate-inputs
Interactively generate all required EUREKA problem input files (INSTRUCTION.md, SUBMISSIONFORMAT.md, evaluate.py) and the launch script (run.sh) from a researcher's natural language description, then validate the evaluation pipeline.
prepare-workspace
Validate problem setup, test the evaluation pipeline, and configure the environment before the propose-implement loop begins.
baseline-comparison-audit
Audit whether a paper's baseline comparisons are COMPLETE, FAIR, and SIGNIFICANT: a required recent SOTA baseline is missing while 'best/SOTA' is claimed (HP-MISSING-BASELINE); a baseline is undertuned / given less compute-tuning-data, run at a mismatched config, or the equal-budget ablation-as-baseline is absent…
eval-design-forensics
Audit whether a paper's EVALUATION DESIGN actually measures what it claims and whether its reporting is complete — the validity layer family D (experiment-forensics) cannot reach. Three patterns: train/test leakage means the reported score may not measure generalization (HP-EVAL-LEAKAGE — adopts the Kapoor & Narayanan…
anti-autoresearch
End-to-end substantive-integrity forensic sweep of a research paper (especially autoresearch / AI-Scientist-style output). Orchestrates the whole pipeline: ingest (arxiv-id | pdf | dir → working dir + pdftotext for L0) → /evidence-ledger (artifact manifest + observability level L0/L1/L2 + span-anchored claims.json) →…
experiment-forensics
Audit experiment integrity against the evidence ledger. At L2 (repo + result files present) a fresh cross-model reviewer reads the eval code line-by-line for fake/derived ground truth, score self-normalization, phantom results (a paper number with no backing file/key), dead/uncalled metric code, verified-scope…