* test: add comprehensive recovery tests for agent loop Add recovery test coverage for all edge cases and failure scenarios in the agentic loop, addressing the missing test coverage for recovery use cases. Tool Call Parsing Edge Cases: - Empty tool_result tags - Empty tool_calls arrays - Whitespace-only tool names - Empty string arguments History Management: - Trimming without system prompt - Role ordering consistency after trim - Only system prompt edge case Arguments Parsing: - Invalid JSON string fallback - None arguments handling - Null value handling JSON Extraction: - Empty input handling - Whitespace only input - Multiple JSON objects - JSON arrays Tool Call Value Parsing: - Missing name field - Non-OpenAI format - Empty tool_calls array - Missing tool_calls field fallback - Top-level array format Constants Validation: - MAX_TOOL_ITERATIONS bounds (prevent runaway loops) - MAX_HISTORY_MESSAGES bounds (prevent memory bloat) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * feat(security): Add Phase 1 security features - sandboxing, resource limits, audit logging Phase 1 security enhancements with zero impact on the quick setup wizard: - ✅ Pluggable sandbox trait system (traits.rs) - ✅ Landlock sandbox support (Linux kernel 5.13+) - ✅ Firejail sandbox support (Linux user-space) - ✅ Bubblewrap sandbox support (Linux/macOS user namespaces) - ✅ Docker sandbox support (container isolation) - ✅ No-op fallback (application-layer security only) - ✅ Auto-detection logic (detect.rs) - ✅ Audit logging with HMAC signing support (audit.rs) - ✅ SecurityConfig schema (SandboxConfig, ResourceLimitsConfig, AuditConfig) - ✅ Feature-gated implementation (sandbox-landlock, sandbox-bubblewrap) - ✅ 1,265 tests passing Key design principles: - Silent auto-detection: no new prompts in wizard - Graceful degradation: works on all platforms - Feature flags: zero overhead when disabled - Pluggable architecture: swap sandbox backends via config - Backward compatible: existing configs work unchanged Config usage: ```toml [security.sandbox] enabled = false # Explicitly disable backend = "auto" # auto, landlock, firejail, bubblewrap, docker, none [security.resources] max_memory_mb = 512 max_cpu_time_seconds = 60 [security.audit] enabled = true log_path = "audit.log" sign_events = false ``` Security documentation: - docs/sandboxing.md: Sandbox implementation strategies - docs/resource-limits.md: Resource limit approaches - docs/audit-logging.md: Audit logging specification - docs/security-roadmap.md: 3-phase implementation plan - docs/frictionless-security.md: Zero-impact wizard design - docs/agnostic-security.md: Platform/hardware agnostic approach Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
5.5 KiB
ZeroClaw Security Improvement Roadmap
Current State: Strong Foundation
ZeroClaw already has excellent application-layer security:
✅ Command allowlist (not blocklist)
✅ Path traversal protection
✅ Command injection blocking ($(...), backticks, &&, >)
✅ Secret isolation (API keys not leaked to shell)
✅ Rate limiting (20 actions/hour)
✅ Channel authorization (empty = deny all, * = allow all)
✅ Risk classification (Low/Medium/High)
✅ Environment variable sanitization
✅ Forbidden paths blocking
✅ Comprehensive test coverage (1,017 tests)
What's Missing: OS-Level Containment
🔴 No OS-level sandboxing (chroot, containers, namespaces) 🔴 No resource limits (CPU, memory, disk I/O caps) 🔴 No tamper-evident audit logging 🔴 No syscall filtering (seccomp)
Comparison: ZeroClaw vs PicoClaw vs Production Grade
| Feature | PicoClaw | ZeroClaw Now | ZeroClaw + Roadmap | Production Target |
|---|---|---|---|---|
| Binary Size | ~8MB | 3.4MB ✅ | 3.5-4MB | < 5MB |
| RAM Usage | < 10MB | < 5MB ✅ | < 10MB | < 20MB |
| Startup Time | < 1s | < 10ms ✅ | < 50ms | < 100ms |
| Command Allowlist | Unknown | ✅ Yes | ✅ Yes | ✅ Yes |
| Path Blocking | Unknown | ✅ Yes | ✅ Yes | ✅ Yes |
| Injection Protection | Unknown | ✅ Yes | ✅ Yes | ✅ Yes |
| OS Sandbox | No | ❌ No | ✅ Firejail/Landlock | ✅ Container/namespaces |
| Resource Limits | No | ❌ No | ✅ cgroups/Monitor | ✅ Full cgroups |
| Audit Logging | No | ❌ No | ✅ HMAC-signed | ✅ SIEM integration |
| Security Score | C | B+ | A- | A+ |
Implementation Roadmap
Phase 1: Quick Wins (1-2 weeks)
Goal: Address critical gaps with minimal complexity
| Task | File | Effort | Impact |
|---|---|---|---|
| Landlock filesystem sandbox | src/security/landlock.rs |
2 days | High |
| Memory monitoring + OOM kill | src/resources/memory.rs |
1 day | High |
| CPU timeout per command | src/tools/shell.rs |
1 day | High |
| Basic audit logging | src/security/audit.rs |
2 days | Medium |
| Config schema updates | src/config/schema.rs |
1 day | - |
Deliverables:
- Linux: Filesystem access restricted to workspace
- All platforms: Memory/CPU guards against runaway commands
- All platforms: Tamper-evident audit trail
Phase 2: Platform Integration (2-3 weeks)
Goal: Deep OS integration for production-grade isolation
| Task | Effort | Impact |
|---|---|---|
| Firejail auto-detection + wrapping | 3 days | Very High |
| Bubblewrap wrapper for macOS/*nix | 4 days | Very High |
| cgroups v2 systemd integration | 3 days | High |
| seccomp syscall filtering | 5 days | High |
| Audit log query CLI | 2 days | Medium |
Deliverables:
- Linux: Full container-like isolation via Firejail
- macOS: Bubblewrap filesystem isolation
- Linux: cgroups resource enforcement
- Linux: Syscall allowlisting
Phase 3: Production Hardening (1-2 weeks)
Goal: Enterprise security features
| Task | Effort | Impact |
|---|---|---|
| Docker sandbox mode option | 3 days | High |
| Certificate pinning for channels | 2 days | Medium |
| Signed config verification | 2 days | Medium |
| SIEM-compatible audit export | 2 days | Medium |
Security self-test (zeroclaw audit --check) |
1 day | Low |
Deliverables:
- Optional Docker-based execution isolation
- HTTPS certificate pinning for channel webhooks
- Config file signature verification
- JSON/CSV audit export for external analysis
New Config Schema Preview
[security]
level = "strict" # relaxed | default | strict | paranoid
# Sandbox configuration
[security.sandbox]
enabled = true
backend = "auto" # auto | firejail | bubblewrap | landlock | docker | none
# Resource limits
[resources]
max_memory_mb = 512
max_memory_per_command_mb = 128
max_cpu_percent = 50
max_cpu_time_seconds = 60
max_subprocesses = 10
# Audit logging
[security.audit]
enabled = true
log_path = "~/.config/zeroclaw/audit.log"
sign_events = true
max_size_mb = 100
# Autonomy (existing, enhanced)
[autonomy]
level = "supervised" # readonly | supervised | full
allowed_commands = ["git", "ls", "cat", "grep", "find"]
forbidden_paths = ["/etc", "/root", "~/.ssh"]
require_approval_for_medium_risk = true
block_high_risk_commands = true
max_actions_per_hour = 20
CLI Commands Preview
# Security status check
zeroclaw security --check
# → ✓ Sandbox: Firejail active
# → ✓ Audit logging enabled (42 events today)
# → → Resource limits: 512MB mem, 50% CPU
# Audit log queries
zeroclaw audit --user @alice --since 24h
zeroclaw audit --risk high --violations-only
zeroclaw audit --verify-signatures
# Sandbox test
zeroclaw sandbox --test
# → Testing isolation...
# ✓ Cannot read /etc/passwd
# ✓ Cannot access ~/.ssh
# ✓ Can read /workspace
Summary
ZeroClaw is already more secure than PicoClaw with:
- 50% smaller binary (3.4MB vs 8MB)
- 50% less RAM (< 5MB vs < 10MB)
- 100x faster startup (< 10ms vs < 1s)
- Comprehensive security policy engine
- Extensive test coverage
By implementing this roadmap, ZeroClaw becomes:
- Production-grade with OS-level sandboxing
- Resource-aware with memory/CPU guards
- Audit-ready with tamper-evident logging
- Enterprise-ready with configurable security levels
Estimated effort: 4-7 weeks for full implementation Value: Transforms ZeroClaw from "safe for testing" to "safe for production"