Copilot cli pack

Packs/CopilotCLI/README.md

@@ -0,0 +1,308 @@
+---
+name: CopilotCLI
+pack-id: pai-copilotcli-v1.0.0
+version: 1.0.0
+author: danielmiessler
+description: PAI's core skills ported to GitHub Copilot CLI environment — Thinking (7 modes), Research (4 modes), Investigation (OSINT+PI), and Agents (trait composition)
+type: skill
+purpose-type: [thinking, analysis, research, investigation, code-assistance]
+platform: copilot-cli
+dependencies: ["github-copilot-cli"]
+keywords: [thinking, research, investigation, osint, pi, agents, copilot-cli, mcp, task-agents]
+---
+
+# CopilotCLI — PAI Skills for GitHub Copilot CLI
+
+> PAI's proven skill system ported to GitHub Copilot CLI. Seven thinking modes, four research modes, OSINT+PI investigation, and trait-based agents — now accessible via the CLI and MCP infrastructure.
+
+---
+
+## The Problem
+
+GitHub Copilot CLI provides powerful command-line code assistance, but lacks the structured thinking, research, and investigation capabilities that PAI's Claude Code packs provide. When working in the terminal, you lose access to:
+
+- **Structured thinking modes** -- no first principles decomposition, adversarial validation, or scientific diagnosis
+- **Multi-mode research** -- no systematic literature search, market analysis, threat intelligence, or competitive research
+- **Investigation frameworks** -- no OSINT/PI workflows for gathering and analyzing information
+- **Agent traits composition** -- no way to combine expertise specializations into unified agents
+
+The fundamental issue: terminal-based work shouldn't mean losing access to sophisticated analytical thinking frameworks.
+
+---
+
+## The Solution
+
+CopilotCLI ports PAI's core skills to the GitHub Copilot CLI environment using MCP (Model Context Protocol) infrastructure and background task agents:
+
+**Thinking** (7 modes) -- First principles decomposition, iterative depth analysis, creative brainstorming, multi-agent council debates, adversarial red teaming, world threat modeling, and scientific hypothesis testing.
+
+**Research** (4 modes) -- Literature search, market analysis, threat intelligence, and competitive research with systematic methodology.
+
+**Investigation** -- OSINT (Open Source Intelligence) and PI (Private Investigation) workflows for structured information gathering and analysis.
+
+**Agents** -- Trait composition system that combines expertise specializations (Security, Infrastructure, DevOps, Full-Stack, etc.) into unified agents for targeted analysis.
+
+The core difference from Claude Code: instead of native skills, CopilotCLI uses `task(mode="background")` agents with MCP tools, storing configurations in `~/.copilot/pai/` instead of `~/.claude/skills/`.
+
+---
+
+## Installation
+
+This pack is designed for AI-assisted installation. Give this directory to your AI and ask it to install using `INSTALL.md`.
+
+**What is PAI?** See the [PAI Project Overview](https://github.com/danielmiessler/Personal_AI_Infrastructure#what-is-pai).
+
+---
+
+## What's Included
+
+| Component | Path | Purpose |
+|-----------|------|---------|
+| Thinking skill router | `src/Thinking/SKILL.md` | Routes requests to 7 thinking modes |
+| Thinking modes | `src/Thinking/Modes/` | FirstPrinciples, IterativeDepth, BeCreative, Council, RedTeam, WorldThreatModel, Science |
+| Research skill router | `src/Research/SKILL.md` | Routes requests to 4 research modes |
+| Research modes | `src/Research/Modes/` | LiteratureSearch, MarketAnalysis, ThreatIntelligence, CompetitiveResearch |
+| Investigation skill | `src/Investigation/SKILL.md` | OSINT and PI workflows |
+| Investigation workflows | `src/Investigation/Workflows/` | Information gathering, source validation, analysis synthesis |
+| Agents skill | `src/Agents/SKILL.md` | Trait composition and agent instantiation |
+| Agent traits | `src/Agents/Traits/` | Expertise specializations: Security, Infrastructure, DevOps, FullStack, Backend, Frontend, DataScience, etc. |
+| MCP configuration | `mcp/` | Tool definitions for task agents |
+| CLI integration | `cli/` | Command wrappers and shell integration |
+
+**Summary:**
+- **Thinking modes:** 7 (same as Claude Code Thinking pack)
+- **Research modes:** 4 (same as Claude Code Research pack)
+- **Investigation workflows:** OSINT + PI
+- **Agent traits:** 8+ specializations
+- **Dependencies:** GitHub Copilot CLI, Node.js 18+ (optional, for pitcrew MCP server)
+
+---
+
+## Platform Mapping: Claude Code → Copilot CLI
+
+| Claude Code Concept | Copilot CLI Equivalent | Notes |
+|---------------------|----------------------|-------|
+| Native Skills | MCP-based task agents | Use `task(mode="background", agent_type="general-purpose")` |
+| ~\.claude\skills\ | ~/.copilot/pai/ | Configuration and custom agent definitions |
+| Task() workflows | task() with shellId | Background agents with full CLI tool access |
+| Claude Code file access | Workspace file access | Uses cwd and view/edit tools |
+| Session state | Session SQL database | sql tool with session database for state tracking |
+| Prompt engineering | MCP tool definitions | Tool schemas define agent capabilities |
+
+---
+
+## Key Differences from Claude Code
+
+### 1. Agent Invocation
+**Claude Code:**
+```
+Use the Thinking skill to decompose this problem
+```
+
+**Copilot CLI:**
+```powershell
+# Via CLI
+copilot think --mode first-principles "decompose this problem"
+
+# Via API (background agent)
+task(mode="background", agent_type="general-purpose", name="think", prompt="decompose this...")
+```
+
+### 2. Tool Access
+**Claude Code:** Native skills have built-in Claude tools (file access, web search, knowledge bases)
+
+**Copilot CLI:** MCP tools (grep, view, powershell, web_fetch, bluebird tools for Azure DevOps, etc.)
+
+### 3. Configuration Location
+**Claude Code:** `~/.claude/skills/`
+
+**Copilot CLI:** `~/.copilot/pai/`
+
+### 4. Persistent State
+**Claude Code:** Skill-managed state files
+
+**Copilot CLI:** SQL session database via `sql()` tool
+
+---
+
+## Prerequisites
+
+- **GitHub Copilot CLI** -- installed and authenticated
+- **Node.js 18+** -- optional, required only for pitcrew MCP server (advanced usage)
+- **Git** -- for version control integration
+- **~/.copilot/pai/** -- directory will be created during installation
+
+---
+
+## Example Workflows
+
+### Terminal-Based Thinking
+
+```
+User: copilot think --mode first-principles "Redesign our auth system"
+
+Response:
+Launching background agent: FirstPrinciples thinking mode...
+
+[Identifies claimed truths about auth architecture]
+[Traces each to fundamental axioms]
+[Separates verified truths from inherited assumptions]
+[Lists reconstruction opportunities]
+
+Key finding: "All users need same auth flow" is an inherited assumption.
+Reality: Different user roles have fundamentally different trust requirements.
+Reconstruction: Design separate auth paths for admin, service, and user tiers.
+```
+
+### Code Investigation with OSINT
+
+```
+User: copilot investigate --mode osint "Research security implications of dependency X"
+
+Response:
+Launching Investigation agent...
+
+[Gathers dependency metadata from npm registry]
+[Analyzes GitHub repository history and issues]
+[Searches threat intelligence databases]
+[Checks known CVE records]
+[Examines maintainer reputation and activity]
+
+Finding: 2 unpatched CVEs, maintenance stalled for 8 months
+Risk level: HIGH
+Recommendation: Migrate to maintained alternative
+```
+
+### Agent Trait Composition
+
+```
+User: copilot agent --traits Security,Infrastructure "Analyze this infrastructure config for security gaps"
+
+Response:
+Instantiating composite agent: Security + Infrastructure specialization...
+
+[Security trait: threat modeling, attack surface analysis]
+[Infrastructure trait: architecture review, deployment patterns]
+[Combined analysis with cross-trait validation]
+
+Results:
+- Security: 3 IAM policy vulnerabilities
+- Infrastructure: 2 redundancy gaps
+- Cross-trait: Auth tokens visible in deployment logs (CRITICAL)
+```
+
+---
+
+## Configuration
+
+### Basic Configuration
+No configuration required. All skills work immediately after installation.
+
+### Optional: MCP Tool Registration
+Register additional MCP tools in `~/.copilot/mcp-config.json`:
+```json
+{
+  "servers": {
+    "pai-pitcrew": {
+      "command": "node",
+      "args": ["~/.copilot/pai/mcp/pitcrew-server.js"]
+    }
+  }
+}
+```
+
+### Optional: Custom Agent Traits
+Add custom traits in `~/.copilot/pai/traits/`:
+```yaml
+# ~/.copilot/pai/traits/CustomTrait.md
+name: CustomTrait
+description: Your expertise description
+keywords: [keyword1, keyword2]
+capabilities:
+  - specific capability 1
+  - specific capability 2
+```
+
+---
+
+## What Makes This Different
+
+Unlike generic CLI assistance, CopilotCLI brings structured analytical thinking to the terminal:
+
+- **Seven complete thinking methodologies**, not just "think step by step"
+- **Four research modes with distinct workflows** (not generic web search)
+- **OSINT/PI investigation workflows** for systematic intelligence gathering
+- **Trait-based agent composition** (not single-mode agents)
+- **Background task execution** with visible transcripts and full CLI tool access
+- **Session state management** via SQL (not ephemeral context)
+- **MCP infrastructure** for extensible tool integration
+
+---
+
+## Invocation Scenarios
+
+| Request | Thinking Mode | Research Mode | Investigation | Agents |
+|---------|---------------|---------------|---------------|--------|
+| "Decompose this from first principles" | FirstPrinciples ✓ | — | — | — |
+| "Search literature on X" | — | LiteratureSearch ✓ | — | — |
+| "Analyze market trends" | — | MarketAnalysis ✓ | — | — |
+| "Investigate a security threat" | — | — | OSINT ✓ | Security ✓ |
+| "Build an agent with Security + DevOps traits" | — | — | — | Traits ✓ |
+| "Red team this architecture" | RedTeam ✓ | — | — | Security ✓ |
+| "Research competitive landscape" | — | CompetitiveResearch ✓ | — | — |
+
+---
+
+## Customization
+
+### Recommended
+No customization needed -- all modes work with sensible defaults.
+
+### Optional
+
+| Customization | Location | Impact |
+|--------------|----------|--------|
+| Thinking modes | `src/Thinking/Modes/*.md` | Adjust mode behavior and workflows |
+| Research sources | `src/Research/Sources.md` | Add custom research data sources |
+| Investigation templates | `src/Investigation/Templates/` | Modify OSINT/PI workflows |
+| Agent traits | `~/.copilot/pai/traits/` | Add custom expertise specializations |
+| MCP tools | `mcp/*.schema.json` | Extend tool definitions |
+
+---
+
+## Credits
+
+- **Original concept:** Daniel Miessler -- PAI system and Claude Code packs
+- **Copilot CLI port:** Enables terminal-based access to PAI's proven analytical frameworks
+- **Inspired by:** The recognition that sophisticated thinking should be accessible everywhere
+
+---
+
+## Related Work
+
+- **PAI Thinking Pack** (Claude Code) -- Original seven thinking modes
+- **PAI Research Pack** (Claude Code) -- Original four research modes
+- **PAI Agents Pack** (Claude Code) -- Trait composition system
+- **GitHub Copilot CLI** -- The platform enabling this capability
+
+---
+
+## Works Well With
+
+- **Blue Bird tools** -- Azure DevOps integration for code history, work items, wiki searches
+- **Kusto tools** -- Data exploration and diagnostics
+- **Domain-specific packs** -- Security, Infrastructure, or DevOps packs for specialized analysis
+
+---
+
+## Changelog
+
+### 1.0.0 - 2026-03-20
+- Initial release: PAI skills ported to Copilot CLI
+- Seven thinking modes: FirstPrinciples, IterativeDepth, BeCreative, Council, RedTeam, WorldThreatModel, Science
+- Four research modes: LiteratureSearch, MarketAnalysis, ThreatIntelligence, CompetitiveResearch
+- Investigation skills: OSINT and PI workflows
+- Agents: Trait composition system with 8+ specializations
+- MCP infrastructure for extensible tool integration
+- Session state management via SQL database
+- Background task execution with full CLI tool access