Quick reference for all AI coding tools
claude "task" — run a taskclaude -p "task" — print mode: one response to stdout, then exit--resume — resume a past session (-c = most recent)--from-pr 42 — load PR context--effort high — set reasoning effort (low to max)--resume ID --fork-session — branch a session/plan — preview changes before execution/batch — split work into parallel worktree agents/loop — rerun a prompt on a recurring interval/debug — enable debug logging for the session/simplify — 4-agent cleanup pass (applies fixes)/compact — summarize history (auto-runs near the context limit)/clear — start a fresh conversation/cost — show token usage & costs/context — visualize context usagedefault — ask before each changeacceptEdits — auto-accept file editsplan — show plan, no executionauto — classify & decide automaticallybypassPermissions — unrestricted ⚠️CLAUDE.md/clear mid-implementation/explain — explain selected code/fix — fix code or linting error/tests — generate tests/doc — generate documentation/new — scaffold new project / file/plan — create implementation plan/startDebugging — generate launch.json/compact — summarize context#file — specific file#folder — folder & its files#codebase — semantic search across workspace#selection — current editor selection#changes — source control diff#fetch — content from a URL@github — repos, issues, PRs, code search@terminal — shell commands & output@vscode — settings, extensions, APIs@filename.ts — specific file (<200 lines)@src/folder/ — folder for patterns@codebase — full repo index searchAGENTS.md — project instructions (also see Always Apply rules)@web "query" — live web search#Context7 — versioned library docs (MCP).cursor/rules/*.mdc — persistent rulesAGENTS.md — project instructions (read by Cursor agent).cursorignore — exclude from indexing.cursor/mcp.json — MCP server config