latitude-llm
Agent observability platform that finds and fixes agent failures, with an optional Jev preclassifier for conversation checks.
Overview
Latitude is an open-source platform for making AI agents 'self-healing'. One line of telemetry captures traces, multi-turn sessions and tool calls; failing traces are auto-grouped into tracked signals with status, size and trend; and Latitude can dispatch coding agents such as Claude Code or Cursor with context, sample traces and a deep link to write a fix and open a PR. It offers an MCP server, semantic search, annotations, flaggers and OpenTelemetry ingest.
Per the listing, Latitude includes an optional Jev preclassifier for conversation checks and records which conversations were selected. The README excerpt describes the platform generally and does not detail the preclassifier, so its questions and thresholds are not confirmed here. It illustrates Jev used as a cheap first-pass filter in front of more expensive evaluation.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Yes / no | Does this conversation need a deeper check (preclassification)?(inferred) | — |
How it works
- Telemetry captures agent traces and sessions
- Optional Jev preclassifier screens conversations
- Selected conversations are checked and grouped into signals
- Coding agents are dispatched to fix recurring failures
Strengths
- Cheap first-pass filter before deeper checks
- Selection records for auditability
- Full observability platform around it
Caveats
- Preclassifier details not in the README excerpt
- Optional component
At a glance
- Type
- Project
- Status
- production
- Language
- TypeScript
- Runtime
- web platform with SDKs
- Has code
- yes
- Has tests
- yes
- License
- MIT
Use cases and integrations
How curated lists describe it
Latitude includes an optional Jev preclassifier for conversation checks and their selection records.
logicrw/awesome-jev-projects
Frequently asked questions
How does Latitude use Jev?
As an optional preclassifier for conversation checks, per the listing.
What is Latitude?
An observability and auto-fix platform for AI agents.
Is Jev required?
No, the preclassifier is optional.
Related projects
CLI that grades the coding-agent sessions on your disk (Claude Code, Codex, opencode, Gemini CLI) with Jev for cents.
CLIbetaTypeScriptSemantic test matchers for Vitest and Jest that ask Jev whether output satisfies a plain-English expectation and threshold the probability.
ProjectbetaTypeScriptpytest plugin for semantic assertions on LLM output, asking Jev whether plain-English claims hold with a 0.8 probability bar.
PluginbetaPython'Vitest for LLM outputs': CLI running YAML/JSON eval suites on Jev for RAG faithfulness, hallucination and agent safety in CI.
CLIprototypeTypeScriptLearning lab for DeepEval with Jev as the judge: replayed evaluations, adjustable decision rules, release gates and drift alerts.
ProjectprototypePythonOffline CLI that measures whether a classifier's confidence matches reality and turns mistake costs into a human hand-off threshold.
BenchmarkbetaPython