The remote Honeycomb MCP server at https://mcp.honeycomb.io/mcp
was working correctly for several days but has suddenly stopped functioning. All data query operations now return "Invalid or missing environment" errors, despite the server successfully authenticating and returning workspace context. The workspace context claims "MCP tools do not function in Classic environments," but these same tools were working fine until recently.
- Working Period: MCP server functioned correctly for multiple days
- Used Successfully: DLQ alarm investigations, production data queries, worker analysis
- Current State: All data queries fail with environment errors
- Basic Functions: Authentication and workspace context still work
From ~/.claude.json
:
"honeycomb": {
"type": "http",
"url": "https://mcp.honeycomb.io/mcp"
}
- Handled via Claude Code's
/mcp
command - Successfully reports: "Authentication successful. Reconnected to honeycomb"
mcp__honeycomb__get_workspace_context
Returns:
HONEYCOMB WORKSPACE CONTEXT
============================
TEAM INFORMATION
----------------
Name: Intercomops
Slug: intercomops
CURRENT TIME
------------
Epoch: 1755781994
Human: 2025-08-21 13:13:14 UTC
ENVIRONMENTS
------------
MCP tools do not function in Classic environments. Please check https://docs.honeycomb.io/troubleshoot/product-lifecycle/recommended-migrations/migrate-from-honeycomb-classic/ or contact Support for more details.
mcp__honeycomb__run_query \
--environment_slug production \
--dataset_slug intercom-production \
--query_spec '{"calculations": [{"op": "COUNT"}], "time_range": 3600}'
Returns: Invalid or missing environment: production
mcp__honeycomb__run_query \
--environment_slug intercom-production \
--dataset_slug intercom-production \
--query_spec '{"calculations": [{"op": "COUNT"}], "filters": [{"column": "is_dlq_error", "op": "=", "value": true}], "breakdowns": ["sqs_queue_name", "app_stage"], "time_range": 86400, "limit": 10}'
Returns: Invalid or missing environment: intercom-production
Based on our team's usage over the past several days:
{
"environment": "production",
"dataset": "intercom-production",
"calculations": [{"op": "COUNT"}],
"filters": [
{"column": "is_dlq_error", "op": "=", "value": true},
{"column": "sqs_queue_name", "op": "=", "value": "production-queue-name"}
],
"breakdowns": ["error.type", "error.msg", "app_stage", "sqs_source_app_stage"],
"time_range": 86400
}
{
"environment": "production",
"dataset": "intercom-production",
"calculations": [{"op": "COUNT"}],
"filters": [{"column": "app_stage", "op": "contains", "value": "worker"}],
"time_range": 86400
}
The core issue: The MCP server successfully authenticates, retrieves workspace metadata, and returns real data (team name, timestamps), but claims that "MCP tools do not function in Classic environments" while simultaneously demonstrating that MCP tools ARE functioning.
This suggests either:
- Recent change: The workspace was migrated TO Classic (unusual direction)
- Server regression: Recent MCP server update broke environment handling
- Partial functionality: Some tools work, others don't, but error messaging is unclear
- Team blocked: Multiple Intercom engineers cannot use MCP for observability automation
- Automation broken: DLQ alarm review workflows non-functional
- Inconsistent behavior: Working authentication/metadata vs failing data queries creates confusion
- Team: Intercomops
- Workspace Type: Appears to be "Classic" (according to server message)
- Previous Status: Unknown, but MCP queries were successful
- Authentication: Working correctly
- Basic connectivity: Working correctly
Reported: 2025-08-21
Reporter: Micheal Looney (Intercom Infrastructure)
Status: Regression - previously working functionality now broken