Tool discovery & drift | Agenetix Docs

After connecting an MCP server, Agenetix discovers its tools automatically. You can rediscover at any time to pick up changes from the live server.

How discovery works#

Agenetix queries your server for its tool catalog and records each tool. When you rediscover later, Agenetix compares the live catalog against what it previously stored:

MERMAID

flowchart TD
    Discover["Run discovery"] --> Fetch["Query tool catalog"]
    Fetch --> Compare["Compare with known tools"]
    Compare --> New["New tools → added"]
    Compare --> Changed["Changed tools → descriptions updated"]
    Compare --> Same["Unchanged tools → no action"]
    Compare --> Gone["Missing tools → disabled"]

What happens during rediscovery#

Scenario	What Agenetix does
New tool appears	Added to your catalog, enabled by default
Tool description or schema changes	The source description and input schema are updated
Tool disappears	Disabled but not deleted — preserves log history and agent references
Previously removed tool reappears	Re-enabled with any previous settings intact

Tools are never permanently deleted during rediscovery. This ensures your request logs, agent configurations, and audit history remain intact even when the upstream server changes.

Monitoring catalog drift#

The server detail page shows a Catalog Status indicator:

In sync — the live server matches your stored catalog
Out of sync — differences detected since last check
Unreachable — Agenetix could not reach the server

Click Rediscover Tools to refresh the catalog. You will see a summary of what changed: new tools, updated tools, unchanged tools, and removed tools.

Optional: improve agent tool selection#

Connected tools are read-only by default because the upstream MCP server owns the tool contract. However, you can optionally customize the display name and description that agents see — this helps agents select the right tool without changing what the upstream server actually exposes.