Recently Evaluated

A lightweight open-source telemetry collector for agentic systems. It stores events in SQLite, serves a local dashboard, and provides HTTP endpoints to send and retrieve agent events.

A Python runtime for decision systems where models propose and deterministic checks decide, with replayable traces, hard checks, and support for agents, knowledge, and guarded tool calls.

Black-box security and contract testing for MCP servers, with baseline diffing, CI failure policies, and report output for JSON, JUnit, and SARIF.

A detailed technical post about IronBee Express and its browser-agent architecture, showing how Jev is used for fast, low-cost step selection and when an LLM takes over.

End-to-end encrypted backup, sync, and sharing for coding agent sessions across machines and teammates. It supports Claude Code, Codex CLI, OpenCode, and Pi, with zero-access storage and CLI-based enroll/push/pull workflows.

A desktop and web app with 51 interactive labs that teaches how modern AI works through hands-on exercises, lessons in plain or technical language, and optional local or cloud AI guidance.

Self-hosted, multi-zone MCP server for Kubernetes with Rust edge services, Python workers, OAuth-based access control, and deployment support for GKE and EKS.

Project-local MCP server that lets existing agents discover each other, share status, and exchange durable direct messages with explicit acknowledgement.

Recly turns an Apple Watch or Galaxy Watch into a recorder, transcribes locally or with your own API key, and stores files in your Google Drive with no Recly server in the middle.

AI-powered German speaking practice for Goethe and telc B1 oral exams, with live roleplay, timed exam simulation, transcription, and sentence-level feedback.

A demo agent chat app showing a multi-agent workflow with roles like Scout, Researcher, Coder, and Reviewer. The page shows past chats, task orchestration, and an approval step before agents write.

A GitHub repository tracking vulnerabilities credited to the Anthropic research team, with a large visible list of CVEs, related findings, and submission guidance.

Loa is a local-first autonomous coding agent for longer engineering tasks, built around a mutable DAG, persistent task state, and structured planning/execution. The repo shows quick start, installation, and runtime details, plus support for local or self-hosted inference and an optional Docker sandbox.

A desktop app for saving AI-generated HTML pages into a local vault, opening them in a sandboxed viewer, and resurfacing them later with spaced repetition. It also exposes an MCP server so agents can save, search, organize, and schedule pages.

An experimental repository showing an LLM-generated binary workflow: natural-language specs are turned into raw executable bytes, then tested in a loop with examples like a hello world, HTTP server, and Tetris.

A talk transcript and write-up on using LLMs to prioritize vulnerability research at scale, with references to two open-source tools: Slice for build-free static analysis and Raink for O(N) listwise ranking.

Revline is an AI web app for researching TikTok and Instagram posts, breaking down viral hooks and formats, and finding creators that fit a brand. It includes searchable public-post research, evidence-backed matches, creator checks, and AI-assisted next-step ideas for brands and creators.

Self-hosted AI business analyst for querying company spreadsheets, contracts, PDFs, emails, and sample data, with answers that cite source files and run locally with your own model key.

Graphite’s Five Percent research article analyzes writing-style differences in Claude Opus 5.5, compares it with prior Claude versions and GPT-6 Astra, and includes charts, tell counts, and downloadable data.

A GitHub repo for checking recorded agent runs against a TLA+ specification of the surrounding protocol. It models retries, acknowledgements, approvals, budgets, and can validate JSONL runs against the same spec.

A Claude Code skill and plugin that bulk-deletes old scheduled-task runs while keeping the newest N and preserving runs where a user replied.

An arXiv paper describing a governed, local-first memory system for AI agents, with audit trails, role-based access, verified erasure, and fault-injection results. The page also links to associated code and a Zenodo DOI.

Self-hosted cloud coding agent infrastructure for running Claude Code and Codex in per-thread sandboxes with a single Docker Compose setup.

Plannotator documents a workflow for letting AI agents ask questions in markdown-style question blocks and collecting answers asynchronously in one place. The page explains how it works, where it works, and how agents and reviewers exchange responses.