USE THIS when the user explicitly asks for an autonomous end-to-end flow that runs brainstorm → spec → spec review → plan → subagent-driven implementation without stopping for confirmation. Trigger phrases include `auto brainstorm → spec → review → plan → subagent-driven impl`, `auto pipeline ...`, `ทำ auto ตั้งแต่ brainstorm ไปจบ`, `เอา Subagent มา brainstorming, spec, review spec, writing-plans, subagent-driven ต้นจนจบ`, `ต้นจนจบไม่ต้องหยุด`, `run full pipeline autonomously`. Overrides the interactive gates of `superpowers:brainstorming` / `superpowers:writing-plans` (which normally ask the user one question at a time and wait for spec approval) because the user has explicitly granted autonomous run authority. Still keeps the discipline of (a) writing a spec, (b) dispatching a reviewer subagent against it, (c) writing a TDD-shaped plan, (d) dispatching one implementer subagent per task with spec-compliance + code-quality reviews between tasks, (e) calling the `advisor` tool at real blocker points instead of
Consult a Principal Software Engineer / Architect — a fresh-context advisor subagent dispatched on **Fable** (fallback **Opus**) — at real decision points, whatever model runs the current session. Use when: architecture/design tradeoffs aren't resolvable from the code; BLOCKED twice on the same root cause; spec conflicts with empirical reality; about to make a hard-to-reverse structural change (public API, data model, concurrency design, dependency boundary). Trigger phrases: `ask the advisor`, `consult the architect`, `second opinion on this design`, `ปรึกษา advisor`, `ขอความเห็น architect`, `ถาม principal ก่อน`, `ขอ second opinion`, and any time `ny-auto-pipeline` says to call the advisor. The advisor ADVISES; the session model (and ultimately the user) DECIDES. Do NOT use for routine implementation questions, obvious errors, decisions the user already made, or as a substitute for reading the code yourself.
Use when a request contains several independent pieces of work, when work can keep flowing while earlier work is still running, or when you want to stay responsive to new instructions instead of blocking on one long task. Trigger phrases include `do these in parallel`, `while that runs, also…`, `manage this as subagents`, `ทำหลายอย่างพร้อมกัน`, `ระหว่างรอ…ทำอย่างอื่นต่อ`.
PREFER THIS over manual `sed`/`grep -rl … | xargs sed`/multi-file Edit batches whenever you need to rename a symbol across files. `astedit rename <OLD> <NEW>` parses the project with tree-sitter via codegraph, resolves cross-file imports with confidence scores, and rewrites only the references the resolver vouches for. Dry-run by default; pass `--apply` to write. Atomic per-file writes, length-based drift detection with SHA-256 fallback. Trigger BEFORE running `sed -i s/X/Y/g`, `rg -l X | xargs sed`, or chains of `Edit` calls renaming the same identifier. Also use when the user asks "rename X to Y", "เปลี่ยนชื่อ symbol", "rename this struct/fn/class across the project", "เปลี่ยน X เป็น Y ทั้งโปรเจกต์". Also use `astedit rewrite --pattern P --rewrite R` for structural codemods — ast-grep pattern syntax with `$X` and `$$$ARGS` metavars, dry-run by default. Trigger BEFORE running ad-hoc `sed`, hand-coded ast-grep CLI invocations, or chained `Edit` calls that match an AST shape rather than a single identifier. Al
PREFER THIS over ad-hoc `grep`/`rg`/`find` whenever you need to answer "where else is X used", "who calls Y", "what does Z call", or "what would break if I change W". `codegraph` parses the project with tree-sitter, resolves cross-file imports, and returns structured results (file, line, column, kind, confidence, reason) — not text matches. Trigger BEFORE running `grep -rn <symbol>`, `rg <fn_name>`, `grep -A`, `find . | xargs grep`, or similar searches for callers/usages/references. Also use when the user asks "find references to X", "who calls Y", "what does X call", "callers of", "callees of", "impact of", "what breaks if I change", "is this function dead", "หาที่ใช้", "ใครเรียกฟังก์ชันนี้", "ถ้าแก้ X อะไรพัง", "มีที่ไหน call". Every hit carries a `confidence` (`high`/`medium`/`low`) and `reason` so you can downrank uncertain matches. Supports Rust, TypeScript, TSX, JavaScript, Python.
PREFER THIS over ad-hoc `grep`/`find`/`rg`/`ls`/`tree`/`fd` whenever you need to survey a codebase, list source files, or locate a symbol. `codemap` parses the project with tree-sitter and returns structured results (file + kind + line range) instead of text matches — faster to reason about, cheaper to feed back into prompts. Trigger BEFORE running `grep -r`, `find . -name`, `rg <symbol>`, `ls src/**`, or similar commands on source code. Also use when the user asks "what's in this repo", "where is X defined", "show me the structure", "list source files", "find the function named Y", "หาฟังก์ชัน", "โครงสร้าง project", "มีไฟล์อะไรบ้าง", "อยู่ไฟล์ไหน". Supports Rust, TypeScript, TSX, JavaScript, Python.