// Enforces task coverage and prevents drift via a deterministic checklist tool with output verification.
End-to-end GECX/CXAS/CES conversational agent lifecycle -- build agents from requirements (PRD-to-agent), create and run evals (goldens, simulations, tool tests, callback tests), debug failures, and iterate to production quality. Use this skill whenever the user mentions GECX, CXAS, CES, SCRAPI, conversational agents, voice agents, audio agents, agent evals, pushing/pulling/linting agents, or agent instructions/callbacks/tools on the Google Customer Engagement Suite platform.
Migrate Dialogflow CX (DFCX) agents to CXAS (Customer Experience Agent Studio) agents. Use this skill when the user mentions DFCX migration, migrating agents, converting DFCX to CXAS, porting agents, agent migration, or post-migration optimization/consolidation. Four independently runnable scripts: migrate.py (1:1), stage_1.py (variable dedup + consolidation), stage_2.py (instruction state machines + tool mocks + lint + report), stage_3.py (rewires consolidated topology from source dep graph; only needed when stage_1 ran consolidation). State persists between scripts via <target>_ir.json so each can run / re-run / resume independently.
Retrieves non-contained CCAI Insights conversations (losses), uses agent intelligence to cluster them into common failure patterns, and generates a markdown report. Use when you need to analyze failure patterns and build targeted regression/evaluation reports.
Automates the ingestion of customer requirement documents such as diagrams, BRDs, code etc., synthesizes high-fidelity natural transcripts, and compiles them into highly interactive, responsive Critical User Journey (CUJ) reports.
A robust methodology for LLM-based requirements gathering and high-fidelity artifact generation. Employs 'Divide, Conquer, and Verify' tactics using specialized subagents, iterative exhaustion loops, and batched execution to ensure zero data loss.
A two-phase protocol for extracting structure and generating transcripts from customer artifacts.
| name | task-coverage-protocol |
| description | Enforces task coverage and prevents drift via a deterministic checklist tool with output verification. |
This protocol governs how you must handle large, repetitive, or multi-step tasks to prevent context drift and ensure 100% task coverage. You must use a deterministic checklist tool to track progress and verify completion.
When given a task that involves processing multiple items (e.g., "analyze all files in this directory"), follow these steps:
List and Count: Use a tool to list the items recursively and print the total count.
Initialize Checklist: Call the checklist tool to initialize a new checklist:
python3 scripts/manage_checklist.py init --title "My Task Title"
Add Items: Iterate through the listed items and add each one to the checklist using the tool. This ensures the LLM does not skip or summarize items:
python3 scripts/manage_checklist.py add --item "item_name" --output_check_path "path/to/output"
--output_check_path pointing to the file or
directory where output for this item will be saved.Verify Count: The tool will return the number of items added. Verify that this matches your initial count.
Iterate and Execute: Process the items one-by-one (or in batches if delegating to subagents).
Mark Done: When an item is processed and output produced, call the tool to mark it as done:
python3 scripts/manage_checklist.py done --item "item_name"
output_check_path and is not empty. If verification
fails, the tool will error and you must redo the work for that item.Conclusion: When all items are marked done, report completion to the user.
If you batch items and assign them to subagents:
done command on the primary checklist only after the subagent
reports success and you verify their output.