cxas-loss-analysis

Insights Loss Analysis & Report Generator

This skill instructs you (the AI Agent) to retrieve recent conversations from CCAI Insights, isolate escalated/non-contained sessions (losses), analyze their root causes to group them into failure patterns, and write a professional Markdown report.

---

Execution Routine

Follow these steps in exact sequence:

Step 1: Parameter Verification

Verify that the user has provided the following required parameters:

• project_id: GCP Project ID hosting Insights.
• location: Insights location (e.g., us).
• app_id: Target CXAS App ID (e.g., db9ee866-28db-458b-b835-78137c974779).
• output_dir: Directory where the final report and test cases will be saved.

And the following optional parameters if they wish to scope the analysis:

• start_time: RFC 3339 timestamp for start of time period (e.g., 2026-05-20T00:00:00Z).
• end_time: RFC 3339 timestamp for end of time period (e.g., 2026-05-26T23:59:59Z).
• filter: Custom API filter string to apply (overrides the default loss filter -labels.sessionContained="true").
• limit: Maximum conversations to retrieve and process (default: 500).

Step 2: Extract Loss Transcripts

Run the lightweight data-extraction script to dump the loss transcripts into chunked JSON files in your workspace.

Command Template:

python3 -P .agents/skills/cxas-loss-analysis/scripts/fetch_losses.py \
  --project-id "{project_id}" \
  --location "{location}" \
  --app-id "{app_id}" \
  --limit {limit} \
  --output-file "{output_dir}/raw_losses.json" \
  [--start-time "{start_time}"] \
  [--end-time "{end_time}"] \
  [--filter "{filter}"]

Note: Always run python using the virtual environment's executable with the -P flag (e.g., .venv/bin/python -P) to avoid path pollution.

Step 3: Read Transcripts & Summarize Escalations

Use the view_file or other file-reading tools to read the generated {output_dir}/raw_losses.json file. Extract the list of chunks (which contains paths to the chunked JSON files).