Feature Request: Optimize `list_applications` Token Usage

[csz-akuity]

Problem

The list_applications tool returns ALL applications with ALL fields, consuming excessive tokens even with small deployments.

Current limitations:

• Only supports search parameter for partial name matching
• No way to filter by project, labels, repository, or namespace
• Returns heavy fields like managedFields, history, and full resource details

Possible Solutions

1. Add filtering parameters (server-side) - Reduce number of applications returned
2. Strip unnecessary fields (client-side) - Reduce payload size per application
3. Add pagination support (client-side) - Limit results returned to AI in single call

[csz-akuity]

Related PRs:
#52
#56
#60

[geka-evk]

Hi @csz-akuity !
Could you please provide some updates regarding the issue. Thanks!

Asking coz I see lots of similar errors with other tools:

● argocd-mcp - get_application (MCP)(applicationName: "moja")
  ⎿  Error: MCP tool "get_application" response (39567 tokens) exceeds maximum allowed tokens (25000). Please use pagination, filtering, or limit parameters to reduce the response size.


● argocd-mcp - get_application_resource_tree (MCP)(applicationName: "moja")
  ⎿  Error: MCP tool "get_application_resource_tree" response (64355 tokens) exceeds maximum allowed tokens (25000). Please use pagination, filtering, or limit parameters to reduce the response 
     size.

[csz-akuity]

Hi @geka-evk

Thanks for bringing this to my attention. This seems to be due to the amount of info stored in the app details and resource tree for your moja application. The two tools, get_application and get_application_resource_tree, don't currently support filtering or pagination, but we can try to add those in to reduce token usage.

You can currently try to use get_application_managed_resources tool which already has extensive filter support implemented (see server.ts). It supports:

• kind
• namespace
• name
• version
• group
• appNamespace
• project