Add Officer Clancy Budget Guardian for AI agent resource management

[Copilot]

Implements budget and attempt tracking for Ralph's autonomous agent loop to prevent runaway costs and infinite loops.

Changes

New module: ralph/budget_guardian.py

• BudgetGuardian class with attempt tracking, budget enforcement, and escalation callbacks
• Four-level escalation: NORMAL → WARNING → CRITICAL → EXCEEDED
• Returns Result[AttemptResult, BudgetExceededError] for each authorization

Runner integration

• Guardian hooks into iteration loop (opt-in via RALPH_BUDGET_ENABLED=true)
• Authorizes attempts, records outcomes, displays budget summary

Configuration

Variable	Default	Description
RALPH_BUDGET_ENABLED	false	Enable guardian
RALPH_MAX_ATTEMPTS	10	Max attempts before denial
RALPH_BUDGET_LIMIT	100.0	Total budget units
RALPH_COST_PER_ATTEMPT	10.0	Cost per attempt

Usage

from ralph.budget_guardian import BudgetConfig, BudgetGuardian

guardian = BudgetGuardian(BudgetConfig(max_attempts=5, budget_limit=50.0))

result = guardian.authorize_attempt()
if result.is_success():
    # execute work
    guardian.record_success()

Also added

• skills/officer-clancy/SKILL.md - skill documentation
• tests/test_budget_guardian.py - 24 tests
• Justfile commands: ralph-budget, ralph-no-budget, ralph-test-budget

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.