2 Commits

Author	SHA1	Message	Date
freemo	876217d0ca	feat(guardrails): implement Per-Session and Per-Org Cost Budgets ... Implements Forgejo issue #584: three-tier budget hierarchy (per-plan -> per-session -> per-org) with tightest limit winning. Domain models: - BudgetLevel enum (PLAN, SESSION, ORG) - BudgetCheckResult (frozen Pydantic model with exceeded_level, warning) - SessionCostBudget (tracks per-session accumulated cost) - OrgCostAccumulator (tracks per-org accumulated cost) - ThreadSafeOrgCostAccumulator (thread-safe wrapper with snapshot) Application services: - CostBudgetService: manages budget state, enforces hierarchy, emits BUDGET_WARNING (once per session) and BUDGET_EXCEEDED events - AutonomyGuardrailService: extended with associate_plan_with_session, check_budget_hierarchy, record_plan_cost_to_session methods Configuration: - Settings: session_max_cost_usd, org_max_cost_usd, budget_warning_threshold Integration: - Session model: cost_budget field, as_cli_dict includes budget data - DI container: CostBudgetService registered as Singleton - CLI session show: cost budget panel display Tests: - 54 Behave scenarios (features/cost_budgets.feature) - 11 Robot Framework integration tests (robot/cost_budgets.robot) - ASV benchmarks (benchmarks/bench_budget_check.py) All nox stages pass: lint, typecheck, unit_tests, coverage_report (98%). CLOSES #584	2026-03-10 14:18:04 -04:00
CoreRasurae	53455275ba	feat(automation): add autonomy guardrails and audit trail ... Add runtime autonomy constraints (max steps, tool budget, required confirmations) and a structured audit trail for plan execution. New domain models: - AutonomyGuardrails: enforces step limits, tool budgets, and confirmation gates with validators and check methods - GuardrailAuditEntry: records each enforcement event with timestamp, event type, guard name, result, reason, and context - GuardrailAuditTrail: ordered collection of audit entries persisted to plan metadata New service: - AutonomyGuardrailService: high-level service for configuring guardrails per plan, checking constraints, recording audit entries, and serializing/restoring state via plan metadata Tests: - Behave: 69 scenarios covering model validation, step/budget/ confirmation checks, audit trail recording, and service operations - Robot: 8 test cases for autonomy guardrail CLI flag smoke testing - ASV: 6 benchmark suites measuring enforcement overhead Documentation: - Updated docs/reference/automation_profiles.md with guardrail fields, enforcement behavior, audit trail schema, and event type reference ISSUES CLOSED: #204	2026-02-28 22:06:57 +00:00