Instruction file imported from youo1/Orderak.APP (.github/instructions/ai-customizations.instructions.md). Copyright stays with the author.

AI customization instructions

Name: ai-customizations
Author: youo1

Use the current GitHub Copilot formats for agents, skills, path instructions, hooks, MCP servers, repository settings, and plugin manifests.
Keep agent tools least-privileged. Read-only personas must not receive edit tools, and remote or destructive actions still require explicit approval.
Put stable policy in instructions, reusable workflows in skills, personas and tool scopes in agents, and deterministic lifecycle enforcement in hooks.
Keep automatic learning append-only, evidence-backed, deduplicated, and constrained to .github/skills/orderak-agent-improvement/references/learned-guidance.md. Direct structural customization changes still require review.
Never commit tokens, credentials, private endpoints, or production identifiers. Shared MCP variables must use the COPILOT_MCP_ prefix and each server must have an explicit tool allowlist.
Preserve protected authentication, localization, security, and documentation contracts.
Run node .github/skills/orderak-agent-improvement/scripts/record-learning.mjs --check and node .github/skills/orderak-agent-improvement/scripts/validate-customizations.mjs after changes.
Run node --test .github/hooks/scripts/hooks.test.mjs .github/skills/orderak-agent-improvement/scripts/record-learning.test.mjs after learning or hook changes.

{ "ocm": "1", "id": "youo1-orderak-app-ai-customizations-instructions", "kind": "skill", "name": "ai-customizations", "description": "Schema, safety, and validation rules for Orderak AI customizations.", "publisher": "youo1", "version": "1.0.0", "capabilities": { "domains": [ "general" ], "tags": [ "skill-md", "github-instructions" ], "languages": [ "en" ] }, "quality_prior": 0.6, "examples": [ "Schema, safety, and validation rules for Orderak AI customizations." ], "primary": false, "metadata": { "source": { "provider": "github-instructions", "repository": "https://github.com/youo1/Orderak.APP", "path": ".github/instructions/ai-customizations.instructions.md", "ref": "566705c3c22cadbed9a54372fb13522d802087ad", "url": "https://github.com/youo1/Orderak.APP/blob/566705c3c22cadbed9a54372fb13522d802087ad/.github/instructions/ai-customizations.instructions.md", "key": "youo1/Orderak.APP/.github/instructions/ai-customizations.instructions.md" }, "applies_to": ".github/agents/**,.github/skills/**,.github/instructions/**,.github/hooks/**,.github/copilot/**,.github/plugin/**,.github/mcp.json,.github/copilot-instructions.md" }, "instructions": "# AI customization instructions\n\n- Use the current GitHub Copilot formats for agents, skills, path instructions,\n hooks, MCP servers, repository settings, and plugin manifests.\n- Keep agent tools least-privileged. Read-only personas must not receive edit\n tools, and remote or destructive actions still require explicit approval.\n- Put stable policy in instructions, reusable workflows in skills, personas and\n tool scopes in agents, and deterministic lifecycle enforcement in hooks.\n- Keep automatic learning append-only, evidence-backed, deduplicated, and\n constrained to\n `.github/skills/orde", "cost": { "context_tokens": 343 } }

ai-customizations

About

AI customization instructions

Use it

Manifest

Reviews