Currently, Codeburn doesn't accurately compute the MCP usage for Opencode. See the following output from optimize:
─── 4. 2 MCP servers configured but never used ───────── Medium ───
Never called in this period: clickup, figma. Each server
loads ~2000 tokens of tool schema into every session.
Potential savings: ~660.0K tokens (~$0.139)
-- Run this command ────────────────────────────────────
Remove unused servers:
claude mcp remove clickup
claude mcp remove figma
As well as from just using codebun command:
╭─────────────────────────────────────────────────────────────────────────────────────╮
│ MCP Servers │
│ No MCP usage │
╰─────────────────────────────────────────────────────────────────────────────────────╯
Here is a previous session from OpenCode, proving the use of the clickup mcp:
Thinking: The user wants me to look at a ClickUp task. Let me fetch the task details using the ClickUp tools.
⚙ clickup_clickup_get_task [task_id=*-3210]
Expected behavior
Codeburn commands accurately report MCP server usage and recommendations for removing servers.
Currently, Codeburn doesn't accurately compute the MCP usage for Opencode. See the following output from optimize:
As well as from just using
codebuncommand:Here is a previous session from OpenCode, proving the use of the clickup mcp:
Expected behavior
Codeburn commands accurately report MCP server usage and recommendations for removing servers.