feat(o11y): add kilo_user_id to API metrics for usage reconciliation

[kilo-code-bot]

Summary

• kiloUserId was already a required field in ApiMetricsParamsSchema (populated by the caller) but was never written to either the Analytics Engine dataset (O11Y_API_METRICS) or the Cloudflare Pipeline stream (API_METRICS_STREAM).
• This PR adds kiloUserId as blob7 in the Analytics Engine writeDataPoint call and as kilo_user_id in the stream payload, making it available in both AE queries and Snowflake exports.
• The api-metrics-schema.json pipeline schema definition is updated accordingly to reflect the new kilo_user_id: string field.
• The purpose is to enable reconciliation of o11y API metrics records with microdollar usage records by user.

Note: The Cloudflare Pipeline stream schema is immutable. The existing stream must be recreated with the updated schema before the new field appears in R2/Snowflake. Run:

./pipelines/recreate-stream.sh o11y_api_metrics_stream pipelines/api-metrics-schema.json \
  o11y_api_metrics_pipeline o11y_api_metrics_sink

Verification

• Deploy the o11y Worker and send an /ingest/api-metrics request; confirm kilo_user_id appears in AE query results and in the stream batch.
• After recreating the Pipeline stream, verify the field lands in Snowflake.

Visual Changes

N/A

Reviewer Notes

• No schema changes to ApiMetricsParamsSchema were needed — kiloUserId was already required there.
• The AE blobs array ordering is positional, so blob7 = kiloUserId appends without shifting existing fields.
• The stream schema change requires the recreate script to be run in production before the field is persisted.

---

Built for Igor Šćekić by Kilo for Slack

[kilo-code-bot]

Code Review Summary

Status: No Issues Found | Recommendation: Merge

Overview

This is a clean, minimal change that correctly wires up kiloUserId → blob7 (AE) and kilo_user_id (Pipeline stream), consistent with the existing blob ordering and the already-required field in ApiMetricsParamsSchema.

Key points verified:

• kiloUserId is z.string().min(1) — always a non-empty string, no null/undefined risk.
• blob7 appends after blob6 (userByok) without shifting any existing positional blobs.
• kilo_user_id is correctly placed between user_byok and ttfb_ms in api-metrics-schema.json, matching the stream payload field ordering.
• The operational requirement to recreate the Pipeline stream before the field lands in R2/Snowflake is clearly documented in both the PR description and the inline comment in o11y-analytics.ts.
• required: true in the pipeline schema is appropriate since the Zod schema guarantees the field is always present.
• New commit (incremental): Adds matching test assertions for blob7 = 'user_123' in the AE blobs array and kilo_user_id: 'user_123' in the pipeline payload — makeValidApiMetricsBody() already included kiloUserId: 'user_123', so no fixture changes were needed. Coverage is complete and correct.

Files Reviewed (3 files)

• services/o11y/pipelines/api-metrics-schema.json
• services/o11y/src/o11y-analytics.ts
• services/o11y/test/index.spec.ts

---

_{Reviewed by claude-sonnet-4.6 · 166,553 tokens}

_{Review guidance: REVIEW.md from base branch main}