Synthetic telemetry shaped like real Claude Code + claude-otel collector output.

Use for:

• Developing and unit-testing scripts/cotel-panther-export.py (the parser must handle the resourceLogs/scopeLogs/logRecords shape, _val wrappers, event bodies like claude_code.user_prompt, claude_code.api_request, claude_code.tool_result, etc.).
• Verifying the JSONL session output (single claude_code_session record per file with embedded turns) matches the spec in the design.

Identifiers (user@example.com, UUIDs, paths) are fictional. Prompt text is representative only.

Files:

• logs-sample.jsonl — main input for the panther exporter (OTEL logs + some startup events).
• metrics-sample.jsonl — for commit/pr count extraction (and general metric shape).
• api-body-pair/ — example raw request/response for api-bodies feature (not used by the 0.2.0 Panther pipeline).

To test the exporter locally (from repo root):

python3 scripts/cotel-panther-export.py \
  --logs examples/samples/logs-sample.jsonl \
  --metrics examples/samples/metrics-sample.jsonl \
  --device-cn "test-device.example.com" \
  --min-idle-hours 0 \
  --out-dir /tmp/test-export

See design-0.2.0.md §12 for full verification procedures.