Logion
nicolasmelo1/logion
Developer tooling for a registry that measures whether AI-agent skills and plugins actually work
Logion is open-source developer tooling (SDK, CLI and agent companion) for a registry that publishes reproducible measurements of AI-agent artifacts such as skills, plugins, MCP servers and models. The repository is tagged for Hermes skills among other agent ecosystems.
What Logion does
Logion addresses the gap that skill and plugin hubs publish provenance and popularity but not whether an artifact works. It resolves which exact version an agent installed, whatever the source (npx skills, a plugin manager, Hugging Face or a bare git clone), and attaches evidence to that version. It attributes results and does not replace your installer. This repository is the client surface, meaning the SDK, CLI and agent companion, and not the platform itself.
Answers are shown in labelled layers instead of one blended score: controlled evaluation reproducible from a pinned contract, static evidence such as scanner results, permissions, license and provenance, field evidence reported by real agents above a minimum cohort size, and an explicit nothing yet. A measurement is pinned to an exact version, content digest, contract digest, evaluator version and environment digest, belongs to a model and harness pair, and must reproduce at least twice before it is called a finding.
Key features
- Measurements pinned to an exact artifact version, content digest and contract digest
- Answers split into controlled evaluation, static evidence, field evidence and nothing yet
- Results tied to a model and harness pair
- Works regardless of how the artifact was installed
- Authors contacted before publication, with rebuttals published at the same URL
When to use it
- Checking whether an installed skill or MCP server has been measured against a reproducible contract
- Attaching evidence to an exact artifact version inside an agent workflow
- Reviewing the method and limits behind a published result before trusting it
Who it is for: Developers and agent users who install third-party skills, plugins, MCP servers or models and want evidence of how they behave.
How it fits with Hermes Agent
A general registry tool that covers several agent ecosystems; the repository carries hermes and hermes-skill topics, but the opening part of the README describes no Hermes-specific integration.
Note: The README says this repository is the client tooling rather than the platform, and that Logion is currently the only issuer of its own measurements.
FAQ
What is Logion?
Logion is an open, versioned registry of AI-agent artifacts such as skills, plugins, MCP servers and models, with provenance and evidence attached. This repository holds its open-source SDK, CLI and agent companion.
Does Logion work with Hermes Agent?
The README does not describe a Hermes-specific integration, but the repository is tagged hermes and hermes-skill. It says Logion attaches evidence to an artifact whatever installer or source it came from.
Is Logion a directory or marketplace?
No. The README says Logion is not an index or directory, not a marketplace and not a safety certification. It publishes measurements of how a specific artifact version behaved under a stated method.
Similar dev tools for Hermes Agent
All dev toolsGovernance harness with a gated plan-to-ship pipeline for agent fleets on Claude Code, Codex, Grok or Hermes
huangrichao2020 hermescheckArchitecture and runtime health checks for Hermes Agent checkouts, forks and deployments
0xNyk openclaw-to-hermesMigration tool that moves an OpenClaw setup to Hermes Agent, with dry-run and backup
BULDEE AI Craftsman SuperpowersArchitecture gate for AI coding agents that blocks writes breaking DDD and Clean Architecture rules
vector4wang EdgeGlowmacOS menu bar app that lights up your screen edges while Claude Code or Hermes Agent is working
magicwubiao go-magicGo AI agent framework inspired by Hermes Agent, with a TUI, web dashboard and messaging gateway
Related guides: How to install Hermes Agent · Run multiple Hermes agents with profiles