GV

GeezerVison

The World Wide Web, restored to 800 pixels of glory.

Home|About|Gaming cabinet|Archives|How it works|Starter sites

On this site

  • Sign in
  • ActionsCI/CD and automation
  • CodespacesCloud development environments
  • All Features
  • Documentation
  • Training
  • Certifications
  • Build with Copilot
  • Customize Copilot
  • Integrations
  • Trending repos
  • GitHub Sponsors
  • Developer articles
  • Enterprises

This view

  • Original page
  • New address

GeezerVison is showing github.com — RightRudderAI - Overview

@RightRudderAI RightRudderAI Follow
More
  • Overview
  • Repositories
  • Projects
  • Packages
  • Stars
@RightRudderAI
RightRudderAI
Follow
View RightRudderAI's full-sized avatar

Right Rudder by ERA RightRudderAI

Follow
Catch the step where an AI agent contradicts a decision it already made, and read the functional life a run has left (the expiry read).
  • Embedded Risk Analytics
  • https://embeddedriskanalytics.com
More
  • Overview
  • Repositories
  • Projects
  • Packages
  • Stars
RightRudderAI/README.md

Right Rudder · by Embedded Risk Analytics

Catch the step where an AI agent contradicts a decision it already made, and read how much functional life the run has left.

Long-running agents rename a field, then keep writing the old name. They fix a bug, then reintroduce it. They commit an order, then commit it again. The run reports success, the tests pass, and the contradiction ships. Right Rudder reads the agent's own trace, folds the actions that succeeded into a ledger of committed state, and names the step where the agent broke faith with itself.

The expiry read. Every agent run spoils eventually. From the same trace, Right Rudder reports how many steps a run has left before its committed state contradicts itself, and raises an alarm while a rejected or corrupted action stands in the record. The read moves when the agent acts and stays flat while it only looks, and a wall clock enters nowhere.

pip install right-rudder
right-rudder demo                 # the committed-state read on the bundled example
right-rudder read chatdev.log     # the read on a framework's own log file, format recognised
right-rudder expiry trace.json    # functional life remaining, and the exposure alarm
Repository What it is
right-rudder Adapters and a CLI for the committed-state read and the expiry read, over fifteen trace formats. Start here.
coherence-census Every framework the read has run on, what it caught, and a trace that replays each catch in one command.
langchain-right-rudder The read and the repair as LangChain agent middleware.
right-rudder-openshell The read for agents in NVIDIA OpenShell sandboxes, joined to the platform's own delivery record.
right-rudder-prime-agent The read over Prime Agent sessions, the orchestrator's and its children's.
right-rudder-foreman The read on Foreman runs, beside the Jev verification loop.
jev-coherence-bench A benchmark that sets the committed-state read against a typed verifier.

Right Rudder was called Fathom until October 2026. The old package names, fathom-read and langchain-fathom, now install the new ones.

Send us a trace, get a readout. Research pages and case studies, including the paper behind the read, are at embeddedriskanalytics.com/research. Paper: SSRN 6683578. Contact: [email protected].

Pinned Loading

  1. right-rudder right-rudder Public

    Catch the step where an AI agent contradicts its own earlier decision. CLI and adapters for LangGraph, CrewAI, Letta, OpenInference, DBOS and multi-agent logs, plus an expiry read for long-running …

    Python 1

  2. coherence-census coherence-census Public

    Reproducible coherence failures in real AI agents. One row per framework (LangGraph, CrewAI, Letta, OpenHands, Phoenix, DBOS), each with a trace that replays in one command.

    Python 1

  3. right-rudder-foreman right-rudder-foreman Public

    Coherence read for Foreman, the Codex supervisor. Flags duplicate commits and stale references on the same run Jev verifies.

    Python

  4. right-rudder-openshell right-rudder-openshell Public

    Coherence read for agents in NVIDIA OpenShell sandboxes. Captures model traffic beside OpenShell's delivery record and flags writes the agent believed landed.

    Python

  5. jev-coherence-bench jev-coherence-bench Public

    Benchmark of a committed-state coherence read against Jev, the typed verifier, on agent runs with injected defects. Standard library only.

    Python

  6. langchain-right-rudder langchain-right-rudder Public

    LangChain agent middleware that flags the step where an agent contradicts its own earlier action. Deterministic, no model calls, fits tests and CI.

    Python

Something went wrong, please refresh the page to try again.
If the problem persists, check the GitHub status page or contact support.

Uh oh!

There was an error while loading. Please reload this page.

You are visitor 01402503 — GeezerVison does not store the pages you look up.
Best viewed with Netscape Navigator 4.7 or Internet Explorer 6 at 800×600. No cookies required.