context-add

Your objective is to capture organizational standards into a versioned context file.

Pre-flight Check

Check that .cocoplus/ exists. If not: Output: "CocoPlus not initialized in this directory. Run $pod init to begin." Then stop.

Present Category Menu

Output:

Which context would you like to add or update?

1. approved-models      — Which Cortex models are approved for use
2. quality-thresholds   — Minimum accuracy and performance standards
3. pii-policy           — PII handling rules for AI functions
4. warehouse-policy     — Which warehouses to use for which query types
5. naming-conventions   — Object naming standards for Snowflake objects
6. governance-gates     — Required approval steps before production deployment

Wait for the developer to select a number (1–6).

Map selection to filename:

• 1 → approved-models.md
• 2 → quality-thresholds.md
• 3 → pii-policy.md
• 4 → warehouse-policy.md
• 5 → naming-conventions.md
• 6 → governance-gates.md

Check Existing Content

If .cocoplus/context/<category>.md already exists, read and display it, then ask: "This file already exists. Would you like to update it (add/change entries) or replace it entirely?"

Wait for response before proceeding.

Collect Standards via Dialogue

For the selected category, ask 5–10 targeted questions. Examples per category:

approved-models:

• Which Cortex models are approved for production classification tasks? (e.g., mistral-large2, llama3.1-70b)
• Which models are approved for extraction/NER tasks?
• Are there models that are explicitly blocked? Why?
• Is there a default model to use when none is specified?
• Any cost or latency constraints that restrict model choice?

quality-thresholds:

• What is the minimum classification accuracy required for production? (e.g., 85%)
• Maximum acceptable p95 latency for AI function calls? (e.g., 2000ms)
• Minimum evaluation dataset size before promoting a model?
• Required precision/recall floor for extraction tasks?