v0.2.9

Security Hardening: Auth/Observability boundaries, connector command controls, registry trust policy, transform isolation, integrity checks, and safe tooling controls

• Delivered observability tier hardening with explicit sensitivity split:
  • Public-safe: /openclaw/health
  • Observability token: /openclaw/config, /openclaw/events, /openclaw/events/stream
  • Admin-only: /openclaw/logs/tail, /openclaw/trace/{prompt_id}, /openclaw/secrets/status, /openclaw/security/doctor
• Delivered constrained transform isolation hardening:
  • process-boundary execution via TransformProcessRunner
  • timeout/output caps and network-deny worker posture
  • feature-gated default-off behavior for safer rollout
• Delivered approval/checkpoint integrity hardening:
  • canonical JSON + SHA-256 integrity envelopes
  • tamper detection and fail-closed handling on integrity violations
  • migration-safe loading behavior for legacy persistence files
• Delivered external tooling execution policy:
  • allowlist-driven tool definitions (data/tools_allowlist.json)
  • strict argument validation, bounded timeout/output, and redacted output handling
  • gated by OPENCLAW_ENABLE_EXTERNAL_TOOLS plus admin access policy
• Extended security doctor coverage with wave-2 checks:
  • validates transform isolation posture
  • reports external tooling posture
  • verifies integrity module availability
• Auth-coverage contract tests were updated to include new tool routes and prevent future route-auth drift regressions.
• Added connector command authorization hardening:
  • separates command visibility from command execution privileges
  • centralizes per-command access checks to reduce cross-platform auth drift
  • supports explicit allow-list policy controls for sensitive command classes
  • adds operator-configurable command policy controls via OPENCLAW_COMMAND_OVERRIDES and OPENCLAW_COMMAND_ALLOW_FROM_{PUBLIC|RUN|ADMIN}
• Added registry anti-abuse controls for remote distribution paths:
  • bounded request-rate controls and deduplication windows reduce abuse and accidental hot loops
  • stale anti-abuse state pruning keeps long-running deployments stable
• Added registry preflight and trust-policy hardening:
  • static package safety checks are enforced before activation paths
  • policy-driven signature/trust posture supports audit and strict enforcement modes
  • registry trust mode is operator-controlled via OPENCLAW_REGISTRY_POLICY and preflight verification enforces fail-closed file-path requirements