dd_incidents_list MCP tool returns only old resolved incidents, not active/stable ones

[Kydoimos97]

Bug

The dd_incidents_list MCP tool consistently returns only old resolved incidents rather than the current active and stable incident queue.

Observed behaviour

Calling mcp__puppy-kit__dd_incidents_list with page_size=100 returned 25 incidents — all with status: resolved, the oldest dating back to 2023. The 26 currently ongoing incidents visible in the Datadog UI were not returned at all.

The tool appears to be sorting or filtering by creation date (oldest first) and not honouring any open/active status filter, effectively making it useless for incident triage.

Expected behaviour

The tool should return active and stable incidents — i.e. the incidents currently open in Datadog — not a page of old resolved ones.

Workaround

The CLI works correctly:

uv run puppy incident list --status active --format json --limit 100
uv run puppy incident list --status stable --format json --limit 100

These return the correct 15 active + 11 stable incidents.

Suggested fix

The MCP tool should either:

1. Default to filtering status=active,stable (open incidents only), or
2. Accept a status parameter matching the CLI options (active, stable, resolved), or
3. Sort by modified descending so recent incidents surface first

The CLI --status flag maps to a state:<value> search query fragment — the same approach should work in the MCP tool.

[Kydoimos97]

Fixed by #15. The default sort/filter bug in dd_incidents_list was addressed in an earlier commit (newest-first sort, status filter param added). PR #15 completes the work by adding a dedicated set-status command and dd_incidents_set_status MCP tool so status transitions are no longer silently broken.