Turn a biomedical question into a traceable review of the evidence.
Search PubMed, retrieve available PMC text, import RIS or CSL-JSON libraries, and build an evidence synthesis with recorded screening, source locations and limitations. Keeps retrieved records, inspected reports, underlying studies and supported claims distinct, so the reader can follow how each conclusion was reached.
$5.00 one-off
Preview — package delivery is not available yet
A file-reading agent and Python 3.10+. Bundled direct tools use the standard library.
License and attribution clarified on 25 September 2026; the MIT grants are preserved. Runtime and example bytes are unchanged. Scientific demonstrations and native-host checks retain their 24 September scope. Individually prepared under the owner's publication authorization. Demonstrations are AI-assisted and self-reviewed; no independent human review of the final generated artifacts is claimed. Updates require an explicit release decision.
Automated retrieval covers PubMed/PMC. A systematic review requires an appropriate protocol and additional sources where needed. The retrieval transport requires a POSIX environment; Linux is tested and native Windows is unsupported.
A bounded reading of the VITAL fracture report follows the source from PubMed to PMC, then into a located evidence matrix and qualified prose. One report was inspected; its supplement and participant data were not assessed.
The exact-title query returned a phrase warning and was marked partial. A verified PMID lookup returned one record; metadata and PMC retrieval succeeded. The log preserves both outcomes. This demonstrates known-report retrieval and reading, not a comprehensive topic search.
This synthetic fixture links a preprint and journal version to one declared study. A third report has unresolved cohort overlap. Distinct identifiers do not establish independent samples, and the unresolved relationship stays visible.
The same three-record synthetic library was imported from RIS and CSL-JSON. Both imports retained all three records, including two sharing a DOI and one without an identifier. Every identity remained unverified.
A clearly synthetic ledger fixture completes its declared review accounting while support_ready remains false. The validator checks the recorded structure and attestations; it does not independently read papers or establish semantic truth. The fixture's identifiers and reviewer labels are synthetic.
All 19 bundled ledger tests passed, alongside the two import examples and live PubMed/PMC checks. Codex discovered the renamed package and ran its CLI help successfully. The historical 103-assertion and 48-provider-test claims are not presented as current results.
35 files · 215 KiB unpacked · standard Agent Skills layout
Package delivery is not available yet. These instructions describe installation once you have the ZIP. The compatibility table records the actual test date, environment and scope for each agent.
Installation overview. Follow the steps for your agent and check the compatibility table for the tested scope.
Preview listing: checkout and package delivery are not enabled.
Preview — package delivery is not available yet
The steps below describe installation once you have the ZIP. Checkout and downloads are not enabled.
Linux shell. The archive contains SKILL.md at its root. Refuse an existing destination to protect local edits.
test ! -e .agents/skills/biomedical-literature-evidence && mkdir -p .agents/skills/biomedical-literature-evidence && unzip biomedical-literature-evidence-1.0.0-public.2.zip -d .agents/skills/biomedical-literature-evidenceverified on Codex CLI 0.154.0, 2026-09-24. Linux only. Read the public version and scope; ran the grant selector / biomedical CLI help. This is a discovery and CLI smoke test, not end-to-end scientific validation.
codex exec 'Read the installed $biomedical-literature-evidence skill. State its version and limits, then run: PYTHONDONTWRITEBYTECODE=1 python3 .agents/skills/biomedical-literature-evidence/scripts/pubmed_retrieve.py --help'| Agent | Status | Method | Notes |
|---|---|---|---|
| Codex CLI 0.154.0 | Verified2026-09-24 | Project-local .agents/skills discovery from extracted ZIP | Linux only. Read the public version and scope; ran the grant selector / biomedical CLI help. This is a discovery and CLI smoke test, not end-to-end scientific validation. |
| Claude Code Not tested | Untested | Standard SKILL.md layout | No native invocation of this public edition was tested. |
| Antigravity Not tested | Untested | Explicit file-reading pointer | No native invocation of this public edition was tested. |
| Other agents Not tested | Untested | Explicit file-reading pointer | Requires file access and an appropriate Python runtime. Biomedical retrieval requires POSIX file locking; native Windows is unsupported. |
MIT permits commercial use, modification, redistribution and resale with notices retained. The earlier MIT grant is preserved; retrieved articles and provider data retain their own terms.
The displayed price does not change the license. Terms apply to the identified edition; earlier grants and third-party rights are preserved.
Prepared from the owner-supplied Codex Skills Bundle dated 11 September 2026. Original bundles are preserved; private calibration metadata and Python caches were excluded. This edition was prepared and checked with AI assistance under the owner's publication authorization.
SHA-256 of biomedical-literature-evidence-1.0.0-public.2.zip — verify your download against this hash:
f7c4b43410343d47b417e96cd00a506a83798a00a90f24099c99ea4d64b6a495