context-instructions

Generate or update .github/copilot-instructions.md with enterprise context based on everything configured so far.

Important: Include ALL configured MCP servers

List every MCP server from .mcp.json — both newly installed and pre-existing ones. For each, document:

• What services it covers
• Key tools and when to use them
• Example: workiq covers M365 — "When user asks about emails, meetings, SharePoint docs, or Teams messages, use the workiq MCP server's ask_work_iq tool"

What to generate

Add an ## Enterprise Context section (or update it if it exists) with:

Per-tool blocks

For each configured MCP server/tool:

• MCP server name
• Key tools it provides (list the actual tool names agents should use)
• When to use this tool vs. alternatives
• Team conventions (project keys, space names, URLs)
• Prefer MCP tools over CLI equivalents when available

Skills and agents

Reference any relevant skills or agents:

• "Use the azure-deploy skill for deployments"
• "The context-wizard agent can reconfigure this setup"

Team Intent & Constraints

Retrieve distilled intent from the session store:

SELECT value FROM session_state WHERE key = 'distilled_intent';

Fall back to conversation context if not available. If neither exists, skip this section and note: "No distilled intent available — run /context-distill to generate."

Include the distilled output from Phase 8 (Distill). This is the Intent layer of ISEE — codified so every agent inherits it.

Intent statements

Express what the team values as clear directives:

• "This team prioritises reliability over feature velocity"
• "Security reviews are required for all external-facing changes"

Constraints

Hard rules that agents must always respect:

• "Production deployments require manual approval from a team lead"
• "All PRs must reference a Jira ticket"
• "Secrets must be stored in Azure Key Vault, never in environment variables"

Autonomy boundaries

Render the classified autonomy boundaries from Phase 8 as a structured table: