synthetic-sciences/openscience is an AI workbench that carries out scientific research by reading papers, forming hypotheses, writing and running code, conducting experiments, analyzing results, and preparing reports. Researchers use it for work in machine learning, biology, physics, and chemistry with remote or local models. Catalogue add-ons extend its scientific workflows through skills and instructions.
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 skills add synthetic-sciences/openscience --skill citationsgit clone --depth 1 https://github.com/synthetic-sciences/openscienceWrote 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/synthetic-sciences/openscience/citations)<a href="https://agentmods.dev/skills/synthetic-sciences/openscience/citations"><img src="https://agentmods.dev/badge/skills/synthetic-sciences/openscience/citations/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/synthetic-sciences/openscience/citations"><img src="https://agentmods.dev/badge/skills/synthetic-sciences/openscience/citations.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.00095 | $0.01234 |
| Opus 5.5 | $0.00038 | $0.00494 |
| Sonnet 5.5 | $0.00019 | $0.00247 |
| Haiku 4.5 | $0.00010 | $0.00123 |
Grade A, and why
citations 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 1mo 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 — 90 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Citations
A citation is a claim that a specific document says a specific thing. Language models produce plausible references that do not exist, and reviewers now check. The rule is simple and absolute: nothing enters the bibliography that was not resolved from a live record, and nothing is cited for a claim that was not read.
Non-negotiables
- Resolve before you cite. Every entry comes from a record fetched now: a DOI at Crossref, a work at OpenAlex, an arXiv id, a PMID. Author names, year, title, venue, pages and DOI are copied from that record, never recalled or completed from memory.
- Read before you attribute. Cite a paper for a finding only after reading the passage (abstract at minimum; the result section for numbers). A title is not evidence of what a paper found.
- One canonical identifier per entry. Prefer the DOI; arXiv id for preprints without
one; PMID as a secondary field. The key is
firstauthorYEARkeywordand is stable once used in the text. - Prefer the published version over the preprint when both exist, and say so if the text relied on the preprint's numbers.
- Never pad. A claim needs the one or two sources that establish it, not five that mention the topic.
- Report what could not be resolved instead of leaving a best guess in the file.
Workflow
- Collect the claims that need sources, or the entries to audit.
- Resolve each to a record (
references/apis.mdhas the exact requests). - Read enough of each source to confirm it supports the claim as worded.
- Write the BibTeX from the record; run
scripts/validate_bib.pyon the file. - Return the list of resolved, corrected and unresolved items.
Resolving. Search by title first: Crossref works?query.bibliographic=<title>&rows=3
or OpenAlex works?search=<title>&per-page=3; accept a hit only if the title matches
closely and the first author agrees. With a DOI in hand, fetch
https://api.crossref.org/works/<doi> for the authoritative metadata, or
https://doi.org/<doi> with Accept: application/x-bibtex for a ready entry to tidy. For
arXiv, http://export.arxiv.org/api/query?id_list=<id> returns title, authors and date;
check Crossref for a published version of the same title before citing the preprint.
For biomedical literature, PubMed E-utilities (esearch then esummary) give the PMID,
DOI and journal.
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 1mo ago Changed · +6 lines 582646932fcc
- 1mo ago First seen · 84 lines · 95 tokens per session scan A d33a892011b2
citations is a skill published in the GitHub repository synthetic-sciences/openscience (4,025 stars, last pushed today), licensed Apache-2.0. It adds 95 tokens to every session and 1,234 once invoked, about $0.0004 per session on Opus 5.5. 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 2026-09-14.
Other skills, from other repositories
songsee
Audio spectrograms/features (mel, chroma, MFCC) via CLI.
arxiv
Search arXiv papers by keyword, author, category, or ID.
research-paper-writing
Write ML papers for NeurIPS/ICML/ICLR: design→submit.
paper-revision-author
Revise independently drafted paper sections into one coherent LaTeX body before the abstract is written.
paper-plot-stub
Plot a results CSV (x, ybaseline, yours) as a two-line matplotlib chart and write a PDF. Demo-only.
knowledge-extension-ecosystem-lsp-code-intelligence
A codebase guide to LSP, the standard protocol behind editor features such as symbol queries, file diagnostics, and language-aware code information.