Agent skill · software engineering · jeremylongshore
openevidence-local-dev-loop
Evaluate OpenEvidence product workflows safely with synthetic cases before any clinical rollout. Use when working with OpenEvidence in a healthcare organization. Trigger with "openevidence local dev loop", "OpenEvidence evaluation", or a matching workflow request.
Why this skill is useful
Adds a complete local development workflow with mock server setup and integration testing for OpenEvidence API that the AI wouldn't generate on its own.
What it needs
Requires npm installed locally. Requires openevidence account access. About 2k tokens when loaded. Last updated 2026-08-07. 2,604 stars on the source repository.
What this skill does
OpenEvidence Synthetic Workflow Evaluation Loop Overview Replace the nonexistent local developer loop with a repeatable browser/app evaluation process. Keep inputs minimal, separate observed facts from assumptions, and leave consequential decisions with the named accountable owner. Prerequisites A clearly bounded workflow, accountable clinical owner, and organizational policy Current first-party OpenEvidence documentation and applicable institution agreements Synthetic or properly authorized minimum-necessary data Tool Discipline Use Read, Glob, and Grep to inspect supplied policies, plans, and evidence. Use WebFetch only for current first-party OpenEvidence documentation. Use Write or Edit only when the user requests a named deliverable with an approved destination. Never expose credentials, PHI, recordings, or unrestricted environment output. Current Contract OpenEvidence does not publish a local runtime, sandbox SDK, or public test API in the audited documentation. Synthetic scenarios are the default evaluation input; real PHI requires the full approved data boundary. Evaluation tests workflow fitness and evidence review, not medical-device validation. Authentication Use only the official OpenEvidence web/mobile sign-in or an institution-approved access path. Do not invent API keys, OAuth clients, SDK credentials, service accounts, or private endpoints. Never ask a user to reveal a password, session token, cookie, or recovery code. Instructions 1. Define feature, user role, expected workflow outcome, unacceptable failure, and clinical reviewer. 2. Create synthetic scenarios spanning routine, ambiguous, conflicting-evidence, and failure-path cases. 3. Read the current guide and record the model, feature, and surface used for each run. 4. Execute manually through the supported product, preserving only de-identified prompts, citations, and observations. 5. Have a qualified reviewer score traceability, applicability, uncertainty, and workflow burden. 6. …
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills jeremylongshore/openevidence-local-dev-loop--a89942