typesafe plugins

by Alberto Leal

Consult Jev for bounded typed second opinions during agent work: compare plans, check claims against supplied evidence, prioritize investigations, and challenge conclusions. Uses Vercel AI Gateway by default with configurable providers; skips when the selected key is absent. Includes a standalone uv-run helper and offline contract tests.

by Alberto Leal

Measure Python complexity per function and per call path with ruff and complexipy; optionally add Jev judgments through configurable providers (Vercel by default). Use when prioritizing refactors, investigating over-layering, or verifying that a refactor improved the whole path. Without the selected provider key, semantic review is skipped and static analysis continues. Tests verify behavior.

by Ghaleb Dweikat

A calibrated context sieve for Claude Code. Every large tool result is judged by a System One model before it enters context; irrelevant blocks are replaced with a summary and a recall key.

by 0x7067

Hands the small judgments in a session to TypeSafe's Jev, a System One model that returns typed answers instead of generating text: what kind of request each prompt is, which model tier a subagent needs, whether an edit breaks a rule already written in your CLAUDE.md or AGENTS.md, and which transcript blocks survive compaction. Kept blocks carry over verbatim — no LLM writes a summary.

by 0x7067

Jev drives a real browser: one goal, an indexed action space, one TypeSafe systemOne decision per step. Exposes the jev_browse MCP tool.

by Falconiere Barbosa

Typed judgments from TypeSafe's Jev model at runtime (skill + REST wrapper) — yes/no probabilities, choices, and scores a script can branch on. Standalone, no dependencies.

by Harshit Badiger

Maps where a codebase makes, avoids, or could make semantic decisions, finds the ones worth a System One model, and rejects Jev where code, an LLM, embeddings, classical ML, or a human fits better. Then validates and implements the survivors.

by Oleg Alshaev

A second opinion on every command, edit and MCP call: Jev typed probabilities, a zero-key allowlist, a local audit log, and a prompt-injection sentinel for Claude Code

by jev-enhanced-plugin contributors

Third-party skills enhanced with advisory TypeSafe Jev judgments: two-axis code review, prose humanizing (en/zh-TW), reply checking, simplification gating, and design-tree grilling, plus a procedure for migrating further skills.

by Alexei Ledenev

Auto-picks the right Claude model for each turn — small models for quick edits, mid-tier for code, frontier for hard problems. Uses Jev to classify each request.