thetahealth
- 12 skills
- 0 followers
- 7 hours ago last updated
- ▌ Mirobody · thetahealthSelf-host Mirobody, the open-source health data engine, and use it from an agent. Use when the user wants to run a personal or family health record on their own machine, bring lab reports and wearables (Apple Health, Garmin, Oura, WHOOP) into one place an AI can read, expose their health data to Claude Code, Codex, Cursor, Claude Desktop or another agent over MCP, or operate a stack that is already running (model key, upgrade, backup, ports). Docker is the only requirement; one model API key runs every surface, or no key with every model on the same machine on llama.cpp.
- ▌ Standardize Health Data · thetahealth bundleTurn raw health data into standardized data, and read a lab report without guessing. A lab report PDF, photo, spreadsheet or CSV becomes readings with LOINC codes and UCUM units; an Apple Health export becomes JSON lines; a symptom or diagnosis in any language becomes an ICPC-3 code; any reading becomes a FHIR Observation. Use whenever health data must be standardized for a pipeline, dataset or app, and whenever a blood test, lab report or checkup result is shared and each row must be identified from the vocabulary rather than from memory. Resolution is offline and needs no key; parsing a document needs one model key.
- ▌ Haenv Bench · thetahealthRun models against an existing question pack and produce a board: offline smoke tests, real models, resumable runs, per-batch archiving, how to read cost, how to read the artifacts (eval/responses/**trace**/cases/batch), and why changing a judge means recomputing, not re-running. Triggers when the user says things like "run an eval batch", "produce a report", "backfill a model", "how much did this batch cost", "show a model's tool-call trace".
- ▌ Haenv Synth · thetahealthGenerate patients and generate questions -- from a set of settings (`job.yaml`) or a piece of case-description text, produce a whole longitudinal trajectory that satisfies those constraints plus a machine-adjudicable gold standard, verifying item by item that "does this question even hold together" before it's handed to a model. Covers two entry points (hand-written job / text extraction), the three kinds of thing that must never be mixed (patient facts / gold standard / generation knobs), whether generating questions costs money (the deterministic tier is free, the LLM tier is cached), and how to read a question blocked by the emission gate. Triggers when the user says things like "generate a batch of questions", "synthesize data", "generate from this case", "question generation got blocked", "generate a patient".
- ▌ Lab Report Walkthrough · thetahealthWalk a person through their lab report — read the original document, organize results by panel, flag out-of-range values against the printed reference ranges, compare with their history, and explain in plain language. Use when the user uploads a lab report (PDF/image) or asks what their blood test results mean.
- ▌
- ▌
- ▌
- ▌ Add Eval Agent · thetahealthScaffold a new EvalAgent plugin (config + implementation + registration).
- ▌ Add Target Agent · thetahealthScaffold a new TargetAgent plugin (config + implementation + registration).
- ▌ Review Architecture · thetahealth审查项目架构健康度 — 检查 GitOps 合规、Plugin 可插拔性、CLI/Web 复用。当用户要求架构审查、代码重构后验证、或新增模块后检查依赖关系时使用。
- ▌ Eslbench Report Analysis · thetahealthUse when analyzing ESL-Bench evaluation reports - extracting scores by difficulty dimension, comparing methods, checking fail rates, token costs, duration per query. Triggers on keywords like eslbench report, benchmark results, group by difficulty.