feat(assisted-query): Add metrics search feature flag and forward options

[isaacwang-sentry]

Summary

• Register gen-ai-explore-metrics-search feature flag to gate the metrics AI search bar independently
• Update SearchAgentStartEndpoint to accept the new flag when strategy is "Metrics"
• Forward the full options dict to Seer instead of extracting only model_name, so strategy-specific context (like metric_context) passes through

Test plan

• Verify the gen-ai-explore-metrics-search flag is registered and exposed to the frontend
• Verify the endpoint accepts requests with strategy: "Metrics" when the new flag is enabled
• Verify the full options dict is forwarded to Seer

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

Autofix Details

Bugbot Autofix prepared a fix for the issue found in the latest run.

• ✅ Fixed: State endpoint missing new metrics feature flag check
  • Added gen-ai-explore-metrics-search flag check to state endpoint to match the start endpoint's feature flag logic, allowing users with only the metrics flag to poll for search agent state.

Or push these changes by commenting:

@cursor push 1df9d6d530

Preview (1df9d6d530)

diff --git a/src/sentry/seer/endpoints/search_agent_state.py b/src/sentry/seer/endpoints/search_agent_state.py
--- a/src/sentry/seer/endpoints/search_agent_state.py
+++ b/src/sentry/seer/endpoints/search_agent_state.py
@@ -81,9 +81,12 @@
                 }
             }
         """
-        if not features.has(
+        has_feature = features.has(
             "organizations:gen-ai-search-agent-translate", organization, actor=request.user
-        ):
+        ) or features.has(
+            "organizations:gen-ai-explore-metrics-search", organization, actor=request.user
+        )
+        if not has_feature:
             return Response(
                 {"detail": "Feature flag not enabled"},
                 status=status.HTTP_403_FORBIDDEN,

_{This Bugbot Autofix run was free. To enable autofix for future PRs, go to the Cursor dashboard.}

[aliu39]

what's the reason for the options param change, is it necessary? We're not using any options besides model_name right?
There's nothing wrong w it per se, but validator could let through a random dict of unused options

[isaacwang-sentry]

what's the reason for the options param change, is it necessary? We're not using any options besides model_name right? There's nothing wrong w it per se, but validator could let through a random dict of unused options

We did this so metric_context could pass through to Seer without the sentry endpoint needing to know about every field, if we want we can explicitly extract model_name + metric_context and only forward those.

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 77da5a6. Configure here.}

[cursor]

State endpoint OR logic defeats global killswitch intent

Medium Severity

The state endpoint uses or to combine the two feature flags, while the start endpoint uses and (per reviewer request) so gen-ai-search-agent-translate acts as a global killswitch. The or in the state endpoint undermines that intent — disabling gen-ai-search-agent-translate won't block state polling for orgs that only have gen-ai-explore-metrics-search enabled.

Additional Locations (1)

• src/sentry/seer/endpoints/search_agent_start.py#L132-L141

 

^{Reviewed by Cursor Bugbot for commit 77da5a6. Configure here.}

[aliu39]

Is there validation on Seer side? If so that's fine, otherwise I'd prefer extracting the fields here