Skip to content

Releases: vrraj/axiolex

Release list

Axiolex - Release 2.0.0

Axiolex - Release 2.0.0 Latest
Latest

Choose a tag to compare

View all tags
@vrraj vrraj released this 09 Sep 01:53
v2.0.0
8bdda5b
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

Release Notes

Version 2.0.0 — Initial Public Release

Overview

Axiolex is a centralized tool discovery and execution platform for AI clients, coding tools, enterprise applications, copilots, and agents.

Clients such as Claude Desktop, Cursor, Codex, enterprise copilots, and custom agents can discover and execute relevant tools without configuring every downstream provider, endpoint, credential, or execution mechanism directly.

Architecture

  • Unified server process: REST and MCP Streamable HTTP are served from a single FastAPI process on port 9700. MCP is available at http://localhost:9700/mcp.
  • Python SDK: lightweight client (pip install axiolex) for namespace access, tool discovery, and execution.
  • @axiolex/mcp-gateway: lightweight Node.js bridge for MCP clients requiring local stdio access. No local Axiolex Python, Redis, or retrieval dependencies are required.

Core Capabilities

  • Unified catalog: MCP tools, A2A agent skills, adapter-backed REST services, and local/internal tools indexed in one searchable catalog.
  • Intent-driven discovery: ranked Top-K tool discovery using BM25S + ColBERT hybrid retrieval, with lexical-only retrieval available by disabling ColBERT.
  • Namespace scoping: single-scope, multi-scope, and full-catalog discovery across business domains.
  • Normalized execution: clients execute a selected tool_id through one Axiolex contract while provider transport, endpoint resolution, authentication, and protocol-specific execution remain server-side.
  • Provider interoperability: MCP Streamable HTTP, MCP stdio, and A2A are supported directly; REST-only systems participate through adapters.
  • Flexible client access: Python SDK, REST / OpenAPI, MCP Streamable HTTP, and the @axiolex/mcp-gateway stdio proxy.
  • Management dashboard: Web UI for provider configuration, namespaces, catalog management, discovery testing, retrieval tuning, credentials, and system status.
  • Reference integrations: native A2A Agent Card discovery and the atlassian_rest_to_mcp Jira adapter for REST-only enterprise systems.

Discovery and Orchestration Boundary

Axiolex ranks the query and namespace scope it receives.

The AI client or application retains:

  • query interpretation and expansion;
  • decomposition of multi-step requests;
  • namespace selection;
  • tool selection;
  • workflow sequencing and reasoning over results.

Axiolex provides:

  • cataloging;
  • namespace-scoped discovery;
  • relevance ranking;
  • normalized tool execution.

This supports workflows such as:

reason → discover → select → execute → reason

MCP Tools

Axiolex exposes three MCP tools to AI clients:

Tool Purpose
list_namespaces List enabled tool domains and namespace descriptions
axiolex_discover_tools Discover ranked tools relevant to a natural-language request
axiolex_execute_tool Execute a discovered tool using its tool_id and arguments

Security

  • Two security boundaries: client access to Axiolex and Axiolex access to downstream providers are handled separately.
  • Client authentication: expected to be enforced at the enterprise deployment boundary using OAuth/OIDC, mTLS, API keys, reverse proxies, API gateways, or service meshes. Axiolex does not enforce client authentication directly in the current release.
  • Server-side provider credentials: downstream credentials are resolved and injected by Axiolex and are not exposed to AI clients or applications.
  • AES-256-GCM encrypted secret storage for provider credentials.
  • Environment-variable-first credential resolution with encrypted-store fallback.
  • Credential redaction from logs, REST payloads, and Redis metadata.
  • Audit metadata: discovery and execution activity can capture query scope, ranked results, relevance scores, tool/provider identifiers, and execution latency.
  • Authorization extensibility: RBAC, delegated user identity, and per-user credential propagation can be layered onto the current security boundaries without changing the core discovery and execution contracts.

Compatibility

  • Python 3.10+
  • BM25S lexical retrieval
  • ColBERT semantic retrieval for hybrid ranking
  • Lexical-only mode available by disabling ColBERT
  • Redis required for shared catalog state
  • Node.js required only for clients using the @axiolex/mcp-gateway stdio proxy

Installation

Axiolex server:

git clone https://github.com/vrraj/axiolex.git && cd axiolex
make install
make start
curl http://localhost:9700/status

Client access:

  • Python SDK: pip install axiolex
  • MCP Streamable HTTP: http://localhost:9700/mcp
  • MCP stdio: npx -y @axiolex/mcp-gateway --endpoint http://localhost:9700/mcp
  • REST API: http://localhost:9700

Links

Assets 2
Loading