pi-typesafe
Advisory Jev sidecar for Pi: four pre-flight hazard checks before bash/write/edit and four diff-quality checks after edits.
Overview
pi-typesafe (twilwa) is a TypeSafe Jev coding-assist sidecar for Pi. It runs four pre-flight hazard checks before bash, write and edit, and four diff-quality checks after successful write and edit, with each phase's questions sharing one System One request. Jev returns typed values and the extension applies thresholds and writes fixed feedback text.
Advisory mode is the default: it records assessments and appends qualifying feedback after execution without blocking tools. Blocking requires explicit opt-in, and shadow mode records assessments without changing model-visible results. Missing credentials or any judge failure leaves tools untouched. Assessments are stored as jev-assessment custom session entries. It is pinned to Pi 0.85.1 and tested with 0.87.0, uses Node 24 for tests, and the README suggests trialling it in a disposable Git repository.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Yes / no | Pre-flight hazard check for this bash/write/edit call(inferred) | yes · no |
| Yes / no | Diff-quality check after a successful edit(inferred) | yes · no |
How it works
- Before bash, write or edit, ask four hazard questions in one request.
- After successful edits, ask four diff-quality questions.
- Apply thresholds and append fixed feedback (advisory).
- Record assessments; block only if opted in.
Strengths
- Non-blocking by default.
- One request per phase.
- Fails open on missing keys or errors.
Caveats
- Advisory feedback can be ignored by the model.
- Specific check questions not listed in the excerpt.
Getting started
npm ci && npm run check && pi -e ./src/extension.ts with TYPESAFE_API_KEY set.At a glance
- Type
- Plugin
- Status
- beta
- Language
- TypeScript
- Runtime
- Pi extension
- Has code
- yes
- Has tests
- yes
Use cases and integrations
How curated lists describe it
Pi coding-agent extension built on the TypeSafe AI System One API (Jev).
hellogumbo/awesome-jev
Frequently asked questions
Does it block tool calls?
Not by default; blocking is opt-in.
What happens without a key?
Tool calls are untouched.
Where are assessments recorded?
In jev-assessment custom session entries.
Related projects
Coding-agent hook that asks Jev one yes/no per project rule on every edit or turn and makes the agent fix any violation.
HookbetaTypeScriptAdvisor for coding agents: git and Stop hooks flag test tampering, secrets and skipped hooks, with optional Jev review findings.
HookbetaFuzzy linter CLI that checks a code tree against plain-language rules in one markdown file, judged by Jev via a swappable interface.
ProjectbetaTypeScriptSemantic code linter that reads each method with its callers and callees and asks decision models about defects, kind, line and severity.
CLIbetaTypeScriptPi extension that checks proposed code edits against repository Markdown rules with Jev, reporting rule text and line ranges.
ProjectbetaTypeScriptRust code-review gate asking Jev small typed questions about functions and tests, failing CI only on rules measured 80% right.
CLIproductionRust