Context compaction

Decide if old agent history is still worth keeping.

Run this demo ▶Get an API key →

The decisions Jev makes

In a single call, Jev evaluates each of these — in parallel, against the same input:

noulstill_useful

Is this old tool call/result still useful for the agent's current work?

returns a calibrated yes/no probability.

choicedecision

How should context compaction handle it?

picks one of these options:

keep — keep the full text verbatim
truncate — keep a short reference, drop the body
drop — remove it entirely

The exact request

This is the real payload behind the live demo — copy it, change the state, and you're building:

{
  "model": "jev-latest",
  "state": "In a long coding-agent session, this old step is still in the context window:\n\n[tool_call] read_file(\"package.json\")\n[tool_result] { name: \"web\", version: \"0.3.1\", dependencies: { next: \"15.1\", react: \"19\" } }\n\nThe agent has since finished all dependency work and is now writing tests.",
  "questions": {
    "still_useful": {
      "type": "noul",
      "instructions": "Is this old tool call/result still useful for the agent's current work?"
    },
    "decision": {
      "type": "choice",
      "instructions": "How should context compaction handle it?",
      "criteria": {
        "keep": "keep the full text verbatim",
        "truncate": "keep a short reference, drop the body",
        "drop": "remove it entirely"
      }
    }
  }
}

Wire it into your code

Read the typed answers and branch in plain code — no parsing. Auto-handle the high-confidence cases and route the uncertain ones to a bigger model or a human. It's one API call and output is free, so ask every question you need at once.

Build your own

Every scenario above is a single API call. Try any of them free in the playground, then get a hosted key to ship it — no waitlist.

Run this demo ▶Get an API key →