blurt 🐔 — show it, say it, get structured work back

People think best out loud with their eyes on the screen. They jump between topics, correct themselves, point at things. Your job: capture that with no friction, then do the tedious part (splitting, writing up, picking and marking screenshots, finding the code, filing) with care. Optimise for the user's flow: never interrupt a recording, ask few questions, make review fast.

Scripts live in scripts/ next to this file; run them with uv run <skill-dir>/scripts/<name>.py ... (each declares its deps; uv installs them on first use). Every script has --help. Always reply in the user's language; write items in the language the user spoke.

0. Setup (first run, or when something fails)

• doctor.py → OS, RAM/GPU, ffmpeg, configured ASR, recommendations, todo list.
• Missing ffmpeg/uv: offer the install command; run it only after the user agrees.
• ASR: if none configured, pick from asr_recommendations + the language the user speaks, tell the user in one line what and why (size, local vs cloud, cost), run its setup. Details: reference/asr.md. Cloud keys are set by the user in their env — never ask them to paste keys in chat.
• record.py build — macOS: compiles the native recorder (Xcode Command Line Tools; else a Tk fallback) and installs the Blurt menu-bar app into ~/Applications. Windows: the Start-menu shortcut appears on first recording. Mention it in one line: "Blurt is also in your Applications — ⌥⇧R records from anywhere".
• record.py test → look at the returned frame yourself. Wallpaper-only/black or ok: false ⇒ Screen Recording permission missing; mic_silent: true ⇒ Microphone permission / wrong mic. On macOS the permission belongs to the app hosting you (Terminal, iTerm, VS Code, Claude, Codex…) and that app must be restarted after granting. macOS 15+ may show an "allow … to bypass the window picker" prompt — the user clicks Allow themselves.
• Optional project config .blurt/config.json: export target, owners/module map, column mapping, custom lenses in .blurt/lenses/. Suggest adding .blurt/sessions/ to .gitignore.

1. Record (or pick up a recording)

From the chat: if the thing to look at is a local dev server/app in this repo, check it is running. Start record.py start in the background (--last-region to reuse the last area) and tell the user in a few lines what happens:

• a dimmed overlay: drag to draw the area (or click a window, F = full screen) → 开始录制 / Enter. Only that area is recorded — tabs, bookmarks and other windows stay private;
• 3-2-1 countdown (click to skip), then a small floating bar: timer · mic level · ⏸ pause · ↺ restart/discard · 完成. ⌥⇧P pause/resume, ⌥⇧S finish (Windows: Alt+Shift). The bar is never in the recording;
• talk naturally, point with the mouse; click 完成 — no need to send another chat message. The task ends with DONE video=… → continue with §2 right away (CANCELLED, exit 2 → acknowledge and stop). From the chat you can also record.py stop | pause | resume | restart.

Waiting on a background step (here and in §5): if your harness wakes you when a background task finishes (Claude Code), end your turn — you'll be notified. If it doesn't (Codex), a finished turn can't be resumed by the child process: keep the turn open and wait on that same tool session with long polls (no busy loop); don't send a final response while it runs.

Existing video (QuickTime, OBS, Loom, phone, a teammate's Blurt recording…): create .blurt/sessions/<timestamp>/, copy it in as recording.<ext> (with events.jsonl / meta.json if it came from Blurt), continue with §2.

Blurt app recordings: the menu-bar app records into the current workspace — ~/Blurt/recordings/<ts>/ by default, or <project>/.blurt/sessions/<ts>/ when a project is bound. record.py inbox lists all of them with processed / reviewed flags. "Process my recordings" → do §2–§4 for each unprocessed one (fan out to subagents if available), then one review per session (or merge into one session dir if the user wants a single list). meta.json has author — keep it on items when several people's recordings are combined.

2. Transcribe

transcribe.py run <session>/recording.mp4 → transcript.json + transcript.txt ([mm:ss.s-mm:ss.s] text). Recorded with system audio (record.py start --system-audio, or the app's menu toggle — calls, product demos), the other side's lines are marked (system) / speaker: "system"; unmarked lines are the user. Speaker bleed can repeat a system line on the mic. For Whisper/API backends pass --prompt with a short glossary (product, page/module names, people) from the repo. Also run frames.py scan <video> (visual-activity index for candidates/sheets). A warning in the output, or a transcript that reads as nonsense (SenseVoice turns unsupported languages such as French into fake English), means the language needs another backend: set it up (reference/asr.md) and re-run.

3. Understand → items

Read the whole transcript first, glance at a few frames, then write <session>/items.json (schema: reference/schema.md). Decide per item what it is — the recording decides, not a mode switch:

kind	when	lens
issue	something in the product to fix or polish	reference/lenses/issue.md
idea	something to build / change / borrow ("这个网站的这里好")	reference/lenses/idea.md
note	an observation / finding / fact worth keeping	reference/lenses/note.md
task	an action item that isn't a product issue	reference/lenses/task.md
custom	a lens in .blurt/lenses/ or ~/.blurt/lenses/ fits, or the user asked for it	that file

Principles — judgement, not rules:

• One item = one thing. People jump around, revisit, correct themselves ("不对，是…"), or say two things in one sentence: merge revisits, drop retracted remarks, split compounds. Pure narration is not an item.
• Fix ASR errors from context (repo vocabulary, what's on screen). Keep quote close to what was said.
• Don't invent. When unsure, write your best guess, set confidence: "low", add a short questions entry.
• Read URLs / product names off the frames into source.
• code_refs (issues, inside a repo): grep the visible text / route / component; list the likely path:lines.
• Give the session a title. For idea-heavy or exploratory recordings write a digest (markdown): themes, the strongest ideas, how they connect, suggested next step — it becomes the review page's Overview.

4. Evidence: frames & clips

Pick 1–3 frames per item that show the point (the bug itself, the part of the page that inspired the idea):

• frames.py sheet <video> --from S --to E -o <session>/sheets/<id>.jpg → ~6 diverse settled candidates with timestamps in one image; --at t1 t2 … for specific moments ("这里 / this / look"). Speech often trails the action.
• Marking the spot — never estimate coordinates from a thumbnail or contact sheet:
  1. frames.py grid <video> --at T -o <session>/sheets/<id>-grid.jpg → labelled 0–1 grid plus the recorded cursor and nearby clicks (events.jsonl — exact). Read box edges off the grid (--crop zooms, labels stay full-frame).
  2. frames.py grab <video> --at T -o <session>/frames/<id>-1.jpg --box x,y,w,h; --ring cursor|click when the user pointed at / clicked it. Tiny detail → second frame with --crop.
  3. Look at every annotated frame before using it. Box not on the point → fix and re-grab.
• Animations / flows / timing: frames.py clip --from --to -o <session>/clips/<id>.mp4 (< 20 MB).
• Many items (> ~12)? Fan out frame work to subagents in batches if your harness has them.

5. Review with the user

Always open the review page — the frames are the point. Run review.py <session> in the background (opens the browser, exits when the user finishes). In the same message:

• ≤ 5 items: the full numbered list (kind · title · owner/module, plus questions); more: a short summary (counts by kind, the headline items, open questions);
• two lines on the page: one item at a time — A keep · X drop · J/K next/prev · Z undo · ? keys; fields editable in place; G list view; V overview (when there's a digest); "完成审核" hands it back. The user may answer in chat or on the page. Wait as in §1 until the review summary prints, then reload items.json (dropped items have status: "deleted", answers are in answer, merged items are gone on purpose and listed in the target's merged_from) and continue with §6. If they reply in chat instead, apply it yourself and close the page (pkill -f "review.py <session>").

Unattended runs (started by the Blurt app via claude -p / codex exec — env BLURT_APP=1 — or any other non-interactive mode): don't ask anything; do §2–§4 and write items.json, then finish with a 2–3 sentence summary. With BLURT_APP=1 don't start the review page — the app opens it when you exit; otherwise start it detached (nohup uv run …/review.py <session> >/dev/null 2>&1 &). Never export to external systems unattended.

6. Export / act

Ask once where things go (remember in .blurt/config.json → export). Typical: issues → the team's bug table; ideas → an ideas board / doc; notes → Markdown; tasks → the user's todo tool.

• Feishu/Lark Bitable → feishu.py export <session> "<table url>" [--kind issue|idea|…] (lark-cli if installed, else app credentials; default columns per kind, creates missing ones, uploads frames + clips, resumable). No table yet → feishu.py create "<name>". Other columns → feishu.py fields then --map. Details: reference/export-feishu.md.
• Markdown + CSV → export_local.py <session> (all kinds; always a good local record).
• GitHub Issues / Linear / Jira / Notion / Obsidian / anything else → reference/export-other.md; use the CLI or MCP tool the user has; map fields by meaning. Confirm destination and count before creating anything remotely; report links afterwards. Then offer the natural next step — for issues "want me to start fixing these?" (you have code_refs), for ideas "want a quick prototype / spec of the top one?".