claude-agent-worktrees — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited claude-agent-worktrees (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Bundled commands use $PLUGIN_ROOT ($env:PLUGIN_ROOT in PowerShell; same path suffix) for the plugin root. Set it once: use the host's plugin-root variable when defined (Claude Code: PLUGIN_ROOT="$CLAUDE_PLUGIN_ROOT"), otherwise the absolute path of this plugin's root directory.
Use this skill for Claude Code session orchestration: background agents, worktrees, tmux/iTerm panes, resumes, PR-linked sessions, forked sessions, session names and IDs, remote control, prompt suggestions, and cloud ultrareview.
Check command surface and current background sessions:
python3 "$PLUGIN_ROOT/scripts/claude_code_inspector.py" --commands agents ultrareview --json
claude agents --jsonUse claude agents --json --all only when completed sessions are relevant and the extra transcript/session metadata is needed.
Use claude agents for active background session state and dispatch defaults:
claude agents --json
claude agents --cwd /path/to/project --json
claude agents --model sonnet --permission-mode plan --add-dir /path/to/extraDo not start or leave long-lived background sessions without a reason and a cleanup plan. Use explicit model, permission mode, settings, plugin-dir, and MCP config when reproducibility matters.
Use these root flags for session continuity:
claude --continue
claude --resume <session-id-or-search>
claude --from-pr <pr-number-or-url>
claude --fork-session --resume <session-id>
claude --session-id <uuid>
claude --name "display-name"Prefer exact session IDs or PR URLs when the user wants a specific conversation. Use search/pickers only for interactive use.
Use worktrees when the task should be isolated from the current working tree:
claude --worktree
claude --worktree feature-name --tmux
claude --worktree feature-name --tmux=classicBefore creating worktrees, inspect git status and branch state. Do not create extra worktrees for tiny one-off commands unless isolation is the point.
Remote control, IDE, and Chrome integration are interactive surfaces:
claude --remote-control
claude --remote-control "session-name"
claude --ide
claude --chrome
claude --no-chromeUse them only when the user asks for those integrations. Report how to stop or exit any interactive session started.
claude ultrareview is cloud-hosted and may consume time or quota. Run it only when requested:
claude ultrareview
claude ultrareview main
claude ultrareview <pr-number-or-url> --json --timeout 30Prefer local review or --print analysis for lightweight checks.
Report the current session/agent/worktree state inspected, exact command used, permission/tool settings, any created session/worktree/pane name, and cleanup or resume instructions when a process remains active.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.