Jev AgentA source-verified directory of 487 Jev projects — browser agents, model routers, SDKs and MCP servers — each one showing the exact decision it hands to Jev. Curated by the Awesome Jev community, mirrored here under MIT.
Give Jev the current state and a few options. Get the Agent’s next step.
Core call · setup in the full example
const nextStep = choice("What next?", { login: null, askUser: null, retry: null });
const { answers } = await client.systemOne({ state: "Login form is ready", questions: { nextStep } });
console.log(answers.nextStep.choice);Node.js 20+ · Set TYPESAFE_API_KEY on your server. This page shows code and makes no model requests.
Prunes old tool calls and outputs for Claude Code while retaining surviving text verbatim.
Separately judges whether a tool call and its full output are still needed; code keeps, truncates or drops them.
Avoids rewriting paths, commands and errors into a new summary.
A Claude Code output-pruning plugin filtering selected Bash output before it reaches the main model.
After length and content checks, Jev judges which chunks to retain while original output is archived.
Short output, errors and recognized structured or source content pass through unchanged.
An X/Twitter filtering extension using local rules by default, with optional Jev checks for unmatched replies.
When experimental AI is enabled, Jev evaluates replies missed by rules and local thresholds decide whether to hide them.
Applies reversible local rules first and adds model judgment only when requested.
A context filter for Claude Code that hides irrelevant tool-output blocks and keeps the original text available for recall.
Jev scores whether Read, Bash and Grep output blocks are needed; local thresholds keep relevant or uncertain blocks.
Reduces visible output while retaining a way to retrieve hidden text.
Retrieve by relevance, not resemblance: filter an AI assistant's memories with TypeSafe's Jev
Jev returns a structured decision for the local program; consult the source for the exact decision policy.
Adds structured choices or scores to the workflow; performance and cost benefits have not been independently verified.
Context-pruning proxy for Claude Code and Codex, employing Jev to prune redundant history while keeping tool protocols intact.
Intercepts request payloads and uses Jev to score historical messages, discarding outdated tool noise before forwarding to upstream LLMs.
Decreases input token counts and time-to-first-token while preserving protocol compliance with client tool calling standards.
Claude Code / Codex / pi plugin that hands agent steps needing no text output to Jev (TypeSafe's judgment model) — measured p50 ~230 ms and ~$0.02 per 1,000 judgments, with typed escalation back to the LLM
Jev returns a structured decision for the local program; consult the source for the exact decision policy.
Adds structured choices or scores to the workflow; performance and cost benefits have not been independently verified.
Local-first Chrome extension that uses Jev to filter sexual and solicitation content in X replies with reversible hidden placeholders.
Scores X reply text with Jev Noul questions for sexual content, solicitation, and spam, then combines probabilities with local thresholds and structural signals.
Combines local rules, caching, concurrency and cost controls, and fail-open handling; it requires no X API and keeps hidden comments recoverable.
Filter any live Twitch chat with Jev: a bring-your-own-key Chrome extension
Jev returns a structured decision for the local program; consult the source for the exact decision policy.
Adds structured choices or scores to the workflow; performance and cost benefits have not been independently verified.
Bare, open-source RLM layer for existing coding agents
Jev receives selected source material and explicit questions. It can rank passages, classify records, assess whether evidence supports a claim and judge whether records match for semantic joins. Its returned probabilities let the RLM compare results and decide what to inspect, combine or explore next.
Adds structured choices or scores to the workflow; performance and cost benefits have not been independently verified.
A Pi extension that prunes stale tool history verbatim and falls back to Pi summary compaction when needed.
Judges whether calls and outputs should survive in the context sent to the model.
Retains the original session file and records pruning decisions.
Adds Jev diagnostic-value, priority and routing signals to OpenTelemetry logs before deeper analysis.
Scores log records and judges whether they merit further model analysis.
Can annotate records while preserving the existing archive path.
An Oh My Pi extension that prunes tool history verbatim and reuses decisions to limit prefix rewrites.
Jev judges tool-call and output utility; the extension truncates selected content with recovery notes.
Remembers pruning decisions and reapplies them across requests.
AlphaOptimizer is an open-source tool from AlphaTales that helps Codex work with large command and tool outputs. Instead of sending a huge log or search result straight into the context window, AlphaOptimizer keeps the useful parts visible, keeps the original output available for a limited time, and uses Jev to help rank what matters when an API key is configured.
Jev returns a structured decision for the local program; consult the source for the exact decision policy.
Adds structured choices or scores to the workflow; performance and cost benefits have not been independently verified.
Verdict-based context compaction for DeepSeek Harness — replaces lossy LLM summaries with fast keep/truncate/drop decisions from jev-latest; everything kept stays verbatim. Port of tamaratran/fast-jev-compaction.
Jev returns a structured decision for the local program; consult the source for the exact decision policy.
Adds structured choices or scores to the workflow; performance and cost benefits have not been independently verified.
Codex plugin for context compaction, using Jev within session lifecycle hooks to preserve essential history while pruning noise.
Assesses past conversation turns with Jev when context limits approach, selecting critical tool outputs and user requirements for retention.
Avoids blanket truncation or lossy re-summarization by selectively preserving exact code locations and decision traces.
Ranks Claude Code skills for the current project to reduce the descriptions loaded by default.
Assesses skills against the project stack, directory, and README before adjusting description visibility.
Keeps relevant descriptions while preserving manual access to other skills.
A pre-compaction advisor for DeepSeek Harness. Runs before the standard `compaction-basic` backend, using TypeSafe JEV to safely prune low-value tool results from model context. Original session events stay in the append-only log; only the model-visible view is replaced with compact markers or archive pointers to reduce context bloat.
Jev returns a structured decision for the local program; consult the source for the exact decision policy.
Adds structured choices or scores to the workflow; performance and cost benefits have not been independently verified.
Fast JEV compaction extension for pi
Jev returns a structured decision for the local program; consult the source for the exact decision policy.
Adds structured choices or scores to the workflow; performance and cost benefits have not been independently verified.
A reversible Pi context filter using Jev to judge whether older messages remain useful.
Scores historical fragments and hides low-scoring content from future requests while preserving session history.
Disabling restores full context; enabling sends selected history to TypeSafe.
A BYOK Chrome extension using Jev to score X posts against personal preferences and adjust their display.
Scores relevance, substance, usefulness, and promotion; local weights and thresholds control highlighting, collapsing, or hiding.
Makes personal feed rules adjustable and display changes reversible.
Automatic Jev context clearing for Pi. Keep the conversation, prune stale tool output, retrieve originals without rerunning commands.
Jev returns a structured decision for the local program; consult the source for the exact decision policy.
Adds structured choices or scores to the workflow; performance and cost benefits have not been independently verified.
Curates Codex handoff context by using Jev to select old tool records while retaining selected text verbatim.
Judges relevance of eligible read-only tool records; local rules protect required content and assemble a handoff packet.
Preserves source references, selection reasons, and original text order in the handoff.
What your last session knew, scored against what this one is doing. MCP server: a per-project ledger written as things happen, recalled per task with TypeSafe's Jev evaluation model via Vercel AI Gateway.
Jev returns a structured decision for the local program; consult the source for the exact decision policy.
Adds structured choices or scores to the workflow; performance and cost benefits have not been independently verified.
A community-maintained directory and radar for Jev, highlighting open-source projects with verified code and clear decision architectures.
Jev is TypeSafe's decision model. This directory groups projects by how they use choices, scores and probability judgements, with implementation links to help you assess fit for your own task.
TypeSafe ↗We prioritize open-source projects with clear Jev implementation evidence. Repositories without verifiable source code are excluded from the main directory.
Directory data and design from Awesome Jev by logicrw, used under the MIT License.
