claude-print-automation

# Claude Print Automation 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="$CLAUDE_PLUGIN_ROOT"`), otherwise the absolute path of this plugin's root directory. Use this skill for non-interactive Claude Code runs: `claude --print`, `-p`, JSON output, stream-json input/output, partial messages, hook event streams, JSON schema validation, budget caps, fallback model, prompt suggestions, no-persistence runs, and CI-style reports. ## Inspect First Verify the installed CLI: ```bash python3 "$PLUGIN_ROOT/scripts/claude_code_inspector.py" --commands "plugin validate" doctor auto-mode --json ``` If the user supplied a binary path, pass `--claude "$CLAUDE_CLI_PATH"`. ## Command Assembly Build commands from these decisions, in order: 1. Trust boundary: confirm the directory is trusted because print mode skips the workspace trust dialog. 2. Permission/tool surface: choose `--permission-mode`, `--tools`, `--allowedTools`, and `--disallowedTools`. 3. Context: add `--add-dir`, explicit files, `--mcp-config`, `--settings`, `--agents`, or `--plugin-dir` only when needed. 4. Prompt input: use an argument for short prompts or stdin for long structured prompts. 5. Output contract: choose `--output-format text|json|stream-json`, `--json-schema`, and partial-message/hook-event streaming. 6. Cost/session: choose `--max-budget-usd`, `--fallba...