[WIP] Complete Phase 1 for agent/skill loaders and TypeScript foundation

[Copilot]

Cloud agent has begun work on create and review an... and will update this pull request as work progresses.

Original prompt

create and review and merge this PR:

https://github.com/stephschofield/beth/compare/main...feat/cli-typescript-foundation?expand=1

Title:

feat(cli): Complete Phase 1 - Agent/Skill Loaders + TypeScript Foundation

Body (copy this):

## Summary

Completes Phase 1 of the CLI TypeScript implementation per the approved architecture plan.

### What's New

**Core Loaders:**
- Agent loader (`src/core/agents/loader.ts`) - parses `.agent.md` files with frontmatter
- Skill loader (`src/core/skills/loader.ts`) - parses `SKILL.md` files with trigger extraction
- Handles GitHub Copilot's `chatagent`/`skill` code fence format

**Types:**
- Full TypeScript types for agent/skill schemas
- `TriggerMap` for skill matching on user input

**Tests:**
- 118 tests passing (33 JS + 85 TS)
- Coverage for loaders, types, path validation, doctor command

**Documentation:**
- Updated CLI-ARCHITECTURE.md for Azure OpenAI (Phase 2)
- Added 3 mermaid diagrams to README (IDEO, Quality Gates, Agent Delegation)
- Updated CHANGELOG with unreleased section

### Phase 2 Preview

Next: Azure OpenAI LLM provider integration (tracked in beads: `beth-nak`)

### Verification

```bash
npm run build      # ✅ Compiles cleanly
npm run test:all   # ✅ 118 tests pass
node bin/cli.js doctor  # ✅ All checks pass

The branch is pushed and ready. Network issues with `gh` are blocking the CLI—you'll need to create the PR manually via that link.The branch is pushed and ready. Network issues with `gh` are blocking the CLI—you'll need to create the PR manually via that link.

<analysis>
[Chronological Review: The conversation began with the user identifying a failing job related to secret scanning using Gitleaks. The user was instructed on how to resolve the failure by reviewing the SARIF report, removing or rotating secrets, and updating the Gitleaks configuration. The user then requested to run the workflows again, followed by a command to delegate the task to a cloud agent.]

[Intent Mapping: 
1. "The failing job is the 'Secret scanning' job..." - User seeks to resolve a failing job related to secret scanning.
2. "Run the workflows again" - User requests to rerun the workflows after addressing the issue.
3. "Delegate: 'Delegate to cloud agent'" - User intends to delegate the task to a cloud agent for execution.]

[Technical Inventory: 
- Tool: Gitleaks - Used for detecting secrets in the repository.
- Configuration: .gitleaks.toml - Suggested for allowing specific patterns to suppress false positives.
- Command: gitleaks detect --redact -v --exit-code=2 --report-format=sarif --report-path=results.sarif --log-level=debug - Command used to run Gitleaks.]

[Code Archaeology: No specific code files or functions were modified in this conversation, but the .gitleaks.toml file was mentioned as a potential addition to the repository.]

[Progress Assessment: The user has identified the issue with the secret scanning job and is in the process of resolving it by reviewing the SARIF report and preparing to rerun the workflows. The task of delegating to a cloud agent is also in progress.]

[Context Validation: All necessary context for continuing the work is captured, including the need to address the Gitleaks findings and the user's intent to rerun workflows and delegate tasks.]

[Recent Commands Analysis: 
- Last Agent Commands: "Run the workflows again" and "Delegate: 'Delegate to cloud agent'".
- Tool Results Summary: No specific tool results were returned as the commands were requests for actions rather than executions of tools.
- Pre-Summary State: The agent was preparing to execute the workflows again and delegate tasks to a cloud agent when the token budget was exceeded.
- Operation Context: These commands were executed to address the failing secret scanning job and to streamline the process by delegating tasks.]

</analysis>

<summary>
1. Conversation Overview:
- Primary Objectives: 
  - "The failing job is the 'Secret scanning' job..." - User aims to resolve the secret scanning job failure.
  - "Run the workflows again" - User requests to rerun the workflows.
  - "Delegate: 'Delegate to cloud agent'" - User intends to delegate the task to a cloud agent.
- Session Context: The conversation focused on resolving a failing job related to secret scanning, with instructions provided for addressing the issue and subsequent requests to rerun workflows and delegate tasks.
- User Intent Evolution: The user initially sought to resolve a technical issue and has since moved towards executing workflows and delegating tasks.

2. Technical Foundation:
- Tool: Gitleaks - Used for detecting secrets in the repository.
- Configuration: .gitleaks.toml - Suggested for allowing specific patterns to suppress false positives.
- Command: gitleaks detect --redact -v --exit-code=2 --report-format=sarif --report-path=results.sarif --log-level=debug - Command used to run Gitleaks.

3. Codebase Statu...

</details>



<!-- START COPILOT CODING AGENT SUFFIX -->

Created from [VS Code](https://code.visualstudio.com/docs/copilot/copilot-coding-agent).

<!-- START COPILOT CODING AGENT TIPS -->
---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our [2 minute survey](https://gh.io/copilot-coding-agent-survey).