feat(phase33): custom skills, workflow recipes, and dynamic slash command autocompletion

[De-pitcher]

Phase 33 Summary: Custom Skills & Workflow Recipes

🎯 Key Capabilities Added

1. Skill Manifest & Frontmatter Parser (\�gentcli.skills.manifest):
   • Zero-dependency parser for YAML frontmatter in \SKILL.md\ files with typed parameters (\string, \int, \loat, \�ool, \enum), default fallbacks, and validation.
   • Template prompt renderer with variable substitution for workspace context (\workspace_dir, \�ctive_branch, \workspace_name) and user arguments.
2. Multi-Tier Skill Discovery (\�gentcli.skills.loader):
   • Prioritized cascading discovery: Project workspace (.agentcli/skills/, .skills/, \skills/) > User global home (~/.agentcli/skills/) > Package built-ins (\�gentcli/skills/builtin/).
3. Execution Engine & Context Grounding (\�gentcli.skills.engine):
   • Context injection with live git branch resolution, argument validation, and execution mode coordination.
4. Skill Runner Subagent (\�gentcli.subagents.skill_runner):
   • Registered as \SubAgentType.SKILL_RUNNER\ handling \list, \info,
     un, and
     eload\ actions with full error isolation.
5. Starter Built-In Skills:
   • \code-review: Architectural and security code review with diff focus and strictness options.
   • \ est-generator: Multi-framework test suite generator (\pytest, \unittest, \�itest, \jest).
   • \security-audit: OWASP and secret scanning audit workflow.
6. Dynamic Slash Command & TUI Integration:
   • Added /skill [list | info | run [args] | reload]\ with autocompletion in interactive CLI and Textual TUI.

🛡️ Quality Gates & Verification

• Unit & Integration Tests: 21/21 Phase 33 tests passed, 90/90 core test suite passed.
• Ruff Linting: 0 errors.
• Mypy Type Checking: 0 errors across all 120 source files.