---

name: loglens description: Query signals pipeline OTLP logs via the loglens Athena-backed CLI. Use when you need to search or inspect signals logs, run opinionated filters (logs/top-errors/schema), or execute raw SQL against the compacted or flat log tables in dev/stage/prod. Triggers include requests mentioning loglens, signals logs, OTLP logs, telemetry alerts, push-server logs, or Athena log queries.

Loglens

Quick start

• Ensure AWS credentials for the target environment. If you are using AWS_PROFILE/SSO credentials directly, add --no-assume-role.
• Run from repo: go run ./cmd/loglens ... (or use build/loglens if already built).
• Confirm environments: loglens envs.
• Run a bounded query (required): --day/--hour or --since/--until.

Example:

CopyAWS_PROFILE=platform-dev AWS_REGION=us-west-2 \
  go run ./cmd/loglens --no-assume-role --env dev logs \
  --day 2026-01-06 --service push-server --severity error --limit 20 --output json

Core workflows

1) Opinionated log search (default compacted)

• loglens logs targets the compacted + deduped table by default.
• Always include a time filter to avoid full scans.
• Use --service, --severity, --contains, and repeatable --filter for column/JSON predicates.
• Add --columns or --include-attributes to inspect specific fields.
• Use --table flat for freshest (non-deduped) data in the raw table.

Examples:

Copy# Compacted (deduped) logs
loglens logs --env dev --day 2026-01-06 --service push-server --severity error --limit 50

# Fresh logs from the flat table
loglens logs --env dev --table flat --day 2026-01-06 --hour 23 --contains "error"

# JSON attribute filter
loglens logs --env dev --day 2026-01-06 --service push-server \
  --filter "attributes.app_id IS NOT NULL"

2) Raw SQL

Use loglens query for schema inspection or complex SQL. Prefer explicit databases if needed.

Copyloglens query --env dev \
  --sql "SELECT event_ts, service_name, severity, body FROM telemetry_alerts_dev.int_otlp_logs_compacted_daily WHERE day=DATE '2026-01-06' AND hour=23 LIMIT 20" \
  --output json

3) Top errors

Copyloglens top-errors --env stage --day 2026-01-06 --limit 20 --trace-samples 2 --output json

4) Schema discovery

Copyloglens schema --env dev --table compacted
loglens schema --env dev --table flat

Data sources and Athena alternative

• Loglens queries AWS Glue tables via Athena.
• Tables:
  • telemetry-parser-db.telemetry_otlp_logs_flat (fresh, not deduped)
  • telemetry_alerts_{env}.int_otlp_logs_compacted_daily (compacted + deduped)
• If you need direct Athena access or Glue catalog exploration, use the $athena-queries skill as an alternative.
• Prod is expected to work the same way, but as of 2026-01-12 there is no data in prod.

References

See references/loglens_reference.md for configuration details, environment mappings, and additional examples.