Files
ottttto_safe-exec/CHANGELOG.md

8.8 KiB

Changelog

All notable changes to SafeExec will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

[0.3.3] - 2026-02-26

Security

  • Enhanced SKILL.md metadata - Added explicit declarations for network, monitoring, and credential requirements
  • Documented Agent Mode - Clearly explained non-interactive execution behavior with full audit logging
  • Added Security & Privacy section - Comprehensive documentation of what SafeExec does and does NOT do
  • Created CLAWDHUB_SECURITY_RESPONSE.md - Detailed response to security review concerns

Changed

  • SKILL.md - Added comprehensive metadata section
    • Declares environment variables: SAFE_EXEC_DISABLE, OPENCLAW_AGENT_CALL, SAFE_EXEC_AUTO_CONFIRM
    • Declares write paths: ~/.openclaw/safe-exec/, ~/.openclaw/safe-exec-audit.log
    • Explicitly states: network=false, monitoring=false, credentials=[]
  • SKILL.md - Added "Security & Privacy" section
    • Clearly documents what SafeExec does and does NOT do
    • Addresses all ClawdHub security review concerns
  • SKILL.md - Enhanced "Agent Mode" section
    • Explains non-interactive execution behavior
    • Documents full audit logging for agent-executed commands
    • Clarifies safety preservation in agent mode

Security Notes

  • No monitoring - Does not read chat sessions or conversation history
  • No network calls - Works entirely locally (except git clone during manual installation)
  • No external notifications - No integration with Feishu, webhooks, or external services
  • No background processes - No cron jobs or persistent monitoring daemons
  • Transparent audit logging - All executions logged with mode label (user_approved / agent_auto)
  • ⚠️ Agent mode preserved - Non-interactive bypass for automation, fully audited

Addressed Issues

This release directly addresses security review concerns from ClawdHub:

  • Declared capabilities: Explicit metadata in SKILL.md
  • Documented behavior: Agent mode clearly explained with safety guarantees
  • Transparency: Comprehensive "Security & Privacy" section

[0.3.2] - 2026-02-26

Security

  • Removed monitoring subsystem - Deleted unified-monitor.sh and all monitoring components
  • Removed external integrations - No more Feishu notifications, GitHub monitoring, or OpenClaw comment checking
  • Simplified project scope - Focused purely on command approval functionality

Removed

  • UNIFIED_MONITOR.md - Unified monitoring system documentation
  • docs/GITHUB_ISSUE_MONITOR.md - GitHub issue monitoring documentation
  • docs/BLOG.md / docs/BLOG_EN.md - Blog posts with notification references
  • docs/CONTRIBUTING.md - Outdated contribution guide
  • docs/FIX_REPORT_v0.1.3.md / docs/FIX_REPORT_v0.2.3.md - Historical fix reports
  • docs/GITHUB_RELEASE_v0.2.0.md - GitHub release documentation
  • docs/GLOBAL_SWITCH_GUIDE.md - Global switch usage guide
  • docs/PROJECT_REPORT.md - Project report
  • docs/PUBLISHING_GUIDE.md - Publishing tool documentation
  • docs/RELEASE_NOTES.md - Release notes
  • docs/RELEASE_v0.2.0.md / docs/RELEASE_v0.2.4.md - Historical release documentation
  • docs/USAGE.md - Usage documentation
  • tools/publish-to-github.sh - GitHub publishing script
  • tools/push-to-github.sh - Git push script
  • tools/release.sh - Release automation script
  • RELEASE_v0.3.2.md - Release documentation
  • UPDATE_NOTES.md - Update notes

Changed

  • README_EN.md - Removed Feishu environment variable configuration

[Unreleased]

[0.2.4] - 2026-02-01

Fixed

  • Non-interactive hang issue: Fixed safe-exec-approve.sh hanging when called by OpenClaw Agent
  • Script now detects non-interactive environments and skips confirmation prompt
  • Added OPENCLAW_AGENT_CALL and SAFE_EXEC_AUTO_CONFIRM environment variable support
  • TTY detection using [[ -t 0 ]] for automatic environment detection

Changed

  • Interactive confirmation is now conditional based on environment
  • Human terminal usage maintains safety confirmation
  • Agent calls automatically bypass confirmation (prevents hanging)

Added

  • FIX_REPORT_v0.2.3.md - Detailed fix report with test results
  • Smart environment detection logic (TTY + environment variables)
  • Visual indicator for non-interactive mode: 🤖 非交互式环境 - 自动跳过确认

Security

  • All security features preserved
  • Danger pattern detection unchanged
  • Risk assessment mechanism unchanged
  • Approval workflow intact
  • Audit logging complete
  • Human users still get confirmation prompt in terminals

Testing

  • Agent call scenario: Pass (no hang, completes in <1s)
  • Environment variable detection: Pass
  • Human terminal usage: Pass (confirmation preserved)
  • Command execution: Pass (successful)
  • Request cleanup: Pass

Backwards Compatibility

  • Fully backwards compatible
  • Existing usage patterns unchanged
  • Human user experience unchanged
  • Agent calls automatically adapt

[0.2.3] - 2026-02-01

Added

  • Context-aware risk assessment: Detect user confirmation keywords
  • Dynamic risk level adjustment based on user intent
  • Customizable confirmation keywords
  • safe-exec-ai-wrapper.sh for AI Agent integration
  • test-context-aware.sh test suite

Changed

  • Risk assessment now considers user context
  • CRITICAL + confirmation → MEDIUM (still requires approval)
  • HIGH + confirmation → LOW (direct execution)
  • MEDIUM + confirmation → LOW (direct execution)

Security

  • CRITICAL operations always require approval
  • All operations logged to audit trail
  • Configurable strictness level

[0.2.0] - 2026-02-01

Added

  • Global on/off switch for SafeExec
  • New commands: --enable, --disable, --status
  • Configuration file safe-exec-rules.json now includes enabled field
  • When disabled, commands execute directly without safety checks
  • Audit log includes bypassed events when disabled

Changed

  • Improved is_enabled() function to handle false correctly
  • Updated status display to show current protection state
  • Enhanced user feedback for enable/disable operations

Fixed

  • Critical Bug: Fixed jq // operator treating false as falsy
  • Now explicitly checks .enabled == true instead of .enabled // true
  • This ensures SafeExec can be properly toggled on/off

Security Note

⚠️ Warning: When SafeExec is disabled, ALL commands execute directly without protection! Only disable in trusted environments.

[0.1.3] - 2026-02-01

Fixed

  • Configuration Fix: Removed incorrect SafeExec plugin configuration
  • SafeExec is now properly configured as a Skill (not a Plugin)
  • Eliminated startup warning logs about missing plugin skill paths
  • Clean separation between Plugin (core extension) and Skill (Agent tool)

Changed

  • Removed ~/.openclaw/extensions/safe-exec/ (incorrect Plugin version)
  • Kept ~/.openclaw/skills/safe-exec/ (correct Skill version)
  • Updated openclaw.json to remove plugin entry for safe-exec

Technical Details

  • Before: SafeExec was registered in plugins.entries.safe-exec
  • After: SafeExec is loaded from skills.load.extraDirs
  • Benefit: Correct architecture, no warning logs, proper Skill behavior

[0.1.2] - 2026-01-31

Added

  • Comprehensive USAGE.md guide (3000+ words)
  • CONTRIBUTING.md with contribution guidelines
  • CHANGELOG.md following Keep a Changelog format
  • release.sh automation script
  • RELEASE_NOTES.md for v0.1.2
  • .github/workflows/test.yml for CI/CD
  • PROJECT_REPORT.md with completion status

Documentation

  • Complete installation guide
  • Usage examples and scenarios
  • Troubleshooting section
  • FAQ (Frequently Asked Questions)
  • Best practices guide

[0.1.1] - 2026-01-31

Added

  • Automatic cleanup of expired approval requests
  • cleanup_expired_requests() function
  • --cleanup flag for manual cleanup
  • Default 5-minute timeout for requests
  • Audit log entries for expiration events

Improved

  • Prevents request database from growing indefinitely
  • Automatic cleanup on every command execution
  • Better security with request expiration

[0.1.0] - 2026-01-31

Added

  • Core risk assessment engine
  • Command interception system
  • Approval workflow
  • Audit logging
  • Integration with OpenClaw
  • Initial documentation

Security

  • 10+ danger pattern detection
  • Fork bomb detection
  • System directory protection
  • Pipe injection prevention

[0.2.0] - Planned

Planned

  • Web UI for approval management
  • Multi-channel notifications (Telegram, Discord)
  • ML-based risk assessment
  • Batch operation support
  • Rate limiting