Organization repository on GitHub
Jev-powered output optimization for Codex, built to keep large tool results concise and usable.
- Forks
- 1
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Integrations & Adapters
AlphaOptimizer
Open-source tool from AlphaTales that sits between Codex and large command outputs: it chunks logs and search results, uses Jev to rank the relevant chunks when a Jev API key is configured, and stores the full output locally for retrieval by artifact ID.
- Stars
- 4
- Language
- TypeScript
- License
- MIT
- Reading time
- 3 min
- Published by
- Community
- Author
- alpha-tales
- Last verified
- Oct 2026
Used inJev for Software & DevTools
- Languages
- TypeScript
- Platforms
- MCP
- Capabilities
- Open sourceRequires API key
Highlights
- In AlphaTales' initial tests, large tool-output payloads were reduced by about 60-65% on average, from as low as 45% to as high as 80%.
- Small outputs, failed outputs, structured data, media, unsupported results and secret-looking content pass through unchanged.
- Jev ranks a bounded set of normal, non-sensitive chunks; the full output is stored in a private local data folder and expires on cleanup rules.
- Installed as a Codex MCP server; its SessionStart and PostToolUse hooks must be trusted in Settings or via /hooks, and changed definitions need re-review.
- The key goes in the MCP server's env table in ~/.codex/config.toml as ALPHAOPTIMIZER_JEV_API_KEY.
Watch out
Copy reviewed against the repo README on 2026-10-15; the captured README was truncated after the setup steps. The 60-65% reduction figures are AlphaTales' own initial tests, and the README explicitly warns they describe context-payload reduction for eligible large outputs, not guaranteed billing savings.
Questions about this entry
Short answers drawn from the linked project or page.
It checks whether an output is worth processing, splits large command and tool outputs into chunks, and sends a small bounded set of non-sensitive chunks to Jev for relevance ranking. Codex then receives a shorter result with selected lines and an artifact ID instead of the whole log.
Jev ranks which chunks matter, so the compacted result keeps the lines that are useful rather than an arbitrary head. Obvious diagnostics such as failures, expected/actual values and important matches are preserved before Jev ranks anything. Jev is used when ALPHAOPTIMIZER_JEV_API_KEY is provided.
Add your Jev API key to the AlphaOptimizer MCP server's env table in Codex's config.toml, then review and trust the SessionStart and PostToolUse hooks — installing the plugin alone does not approve them. Restart Codex and use optimization_status to check configuration and event delivery.
Small outputs, failed outputs, structured data, media, unsupported results and secret-looking content pass through unchanged. The README's flow also returns the original output whenever the gate decides an output should not be processed.
The full supported output is stored in a private local data folder after Jev responds, and Codex can read more detail from it by artifact ID. Stored output expires or is evicted by the configured cleanup rules, so it is not kept forever.
No. The README states the 60-65% figures describe observed context-payload reduction for eligible large outputs in initial AlphaTales tests and are explicitly not guaranteed billing savings. Actual results depend on the shape of the output and the task.
It is an open-source tool from AlphaTales, published under the MIT license. The repository appeared on 2026-09-22 and was last pushed 2026-09-24, so it is a new project; the captured README does not document a roadmap or support commitments.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: Jev Day recap: four times the context and the danger zone
More Jev resources like this
Local tool-routing classifier for coding agents: a gateway and MCP tool that ask Jev which tool to call next, keep your agent's own model and login, and log every decision so you can start by observing.
Jev Integrations & Adapters#community#catalog-import#logicrw
Community
felpsdev/jev-classifierTypeScript
felpsdev/jev-classifierTypeScriptA fork of codex-with-chatgpt that adds TypeSafe Jev plus deterministic rules to decide when Codex hands planning, debugging or review to web ChatGPT and when it switches back. Smart switching is off by default, and with it off the tool behaves exactly like the original.
Jev Integrations & Adapters#community#catalog-import#logicrw
An OpenCode plugin that keeps the session on a cheap sticky parent model so the cache stays warm, and asks Jev whether a turn is hard before escalating through jev_escalate to temporary stronger subagents.
Jev Integrations & Adapters#community#catalog-import#cobanov
A tool-permissions layer for the Vercel AI SDK backed by Jev through the Vercel AI Gateway: a pre-flight approval gate that decides auto-approve, ask-human or block, plus tool-list pruning that drops irrelevant definitions to save context tokens.
Jev Integrations & Adapters#community#catalog-import#logicrw
Verdict-based context compaction for DeepSeek Harness: instead of a lossy summary, jev-latest scores every tool call and result — kept verbatim, truncated to 300 characters or dropped — and falls back to compaction-basic on any failure.
Jev Integrations & Adapters#community#catalog-import#logicrw
A Codex plugin that prepares traceable context handoffs: Jev answers one narrow classification question per old, verified read-only tool pair, and code validates the answers, preserves protected content and builds a JSON or Markdown packet.
Jev Agents & MCP#community#catalog-import#logicrw
From the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.