Routes clinical-data requests to the correct OpenMed workflow skill using deterministic rules.
npx skills add https://github.com/maziyarpanahi/openmed --skill ask-openmed
Please help me install this Agent Skill. Skill: ask-openmed Source: https://github.com/maziyarpanahi/openmed/tree/main/skills/ask-openmed Command: npx skills add https://github.com/maziyarpanahi/openmed --skill ask-openmed
When a request involving clinical data is broad, spans multiple stages, or leaves data sensitivity unclear, it is hard to know which OpenMed workflow to start with. This Skill applies a deterministic routing contract that selects the correct first skill for intake, privacy, extraction, exchange, or verification goals, and enforces a privacy gate before any downstream processing of potentially sensitive content.
Ask the assistant to route your clinical-data goal, for example: determine which OpenMed workflow should handle OCR-ing a scanned lab report and extracting its findings.
High-intent search queries and answers about installing and using this skill.
State your goal and let the routing contract match it against ordered cue tables for intake, privacy, extraction, exchange, and verification. The router returns the selected category, skill identifier, matched rule, and next handoff without processing your data.
If sensitivity is unstated or unclear, the privacy override routes the request to the de-identification skill before any extraction, exchange, or verification. Only explicit safe markers like 'synthetic input' or 'already de-identified' bypass this gate, and negated markers do not count.
No. The router only normalizes the stated goal text and selects a skill identifier. It never inspects the data payload, downloads models, calls services, or makes network requests, and it never echoes input content into logs or reports.
Multi-stage requests are routed one stage at a time in a fixed order: intake, privacy, extraction, exchange, then verification. A later stage named first does not bypass the privacy gate when the data sensitivity is not explicitly safe.
When no cue row matches the normalized goal, the router falls back to the building-with-openmed orientation skill. You can then clarify the goal and sensitivity status and re-route to a focused workflow.
rnavarychConfigures interface engines for EHR interoperability using HL7v2 and CCD/C-CDA documents.
maziyarpanahiExtract clinical notes from FHIR Bulk Data exports and process them through local de-identification and NER.
contentbugvideoeditingEnforce PHI/PII compliance patterns in healthcare applications.
leeoveryGenerate a feature specification markdown file from inline context.
