TypeSafe AI Playground
Rust CLI of Jev experiments: HIPAA-style PHI detection, code-comment review, tone analysis and occupation classification.
Overview
TypeSafe AI Playground is a Rust command-line tool, typesafe-ai, that collects the author's experiments with Jev. Subcommands cover PHI detection, code-comment review, parts-of-speech and load-bearing analysis, interactive tone analysis, and business and occupation classification. It calls Jev through OpenRouter or directly through TypeSafe.
The PHI command is the most documented. It sends a file, a directory of files or stdin to Jev and prints four independent Noul probabilities per file, mirroring the elements of 45 CFR 160.103: identifying information, health condition, healthcare provision and healthcare payment. Files are evaluated concurrently and failures are reported without stopping the scan. Input is sent in full without truncation, so users should be aware that sensitive text goes to the hosted model.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Yes / no | Does the text contain individually identifying information? | yes · no |
| Yes / no | Does the text relate to an individual's health condition? | yes · no |
| Yes / no | Does the text relate to healthcare provision or payment? | yes · no |
How it works
- Run typesafe-ai phi on a file, directory or stdin.
- Each file is sent to Jev in one evaluation.
- Four Noul probabilities are printed per file.
- Users combine them to judge whether the text is PHI.
Strengths
- Metrics map to specific HIPAA definitional elements.
- Concurrent directory scanning with per-file errors.
- Works through OpenRouter or TypeSafe.
Caveats
- Full input goes to a hosted model.
- Described as a playground of experiments.
- Not a compliance tool or legal determination.
Getting started
cargo install --path . --locked; export OPENROUTER_API_KEY or TYPESAFE_API_KEY; typesafe-ai phi notes.txtAt a glance
- Type
- CLI
- Status
- prototype
- Language
- Rust
- Runtime
- CLI
- Has code
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
Rust CLI experiments for health-information screening, comment review, tone analysis and business or occupation classification.
logicrw/awesome-jev-projectsA playground for experiments around Jev, TypeSafe's System One model.
hellogumbo/awesome-jevRust CLI of Jev experiments, including PHI detection, code-comment review, live tone analysis, and occupation and industry classification.
MrJev/awesome-jev
Frequently asked questions
What does the PHI command check?
Four independent elements: identifying information, health condition, healthcare provision and healthcare payment.
Which providers are supported?
OpenRouter (preferred if both keys are set) or TypeSafe directly.
Is it a compliance product?
No, it is an experimental playground.
Related projects
Local-first personal-data inventory tool with optional Jev classification per column, not per row, and dry-run payload review.
ProjectprototypePythonBenchmark of Jev versus rules, local PII models and LLMs at finding personal information: accuracy, leaks, cost and latency.
BenchmarkresearchPythonToolkit for Jev decision contracts: compiled policies, egress controls, tracing, stability checks and certified decision SLOs.
ProjectbetaTypeScriptSpanish compliance triage: clauses extracted deterministically, then Jev labels each obligation, risk or no action.
LibrarybetaPythonPython policy-as-code SDK that runs Noul, Choice and Score rules defined in YAML through Jev, with audit mode and fail-closed options.
LibraryprototypePythonUiPath coded agent for AML alert triage where an LLM reads and writes but Jev makes every red-flag, risk and disposition decision.
AppprototypePython