Summarize Call

Summarize Call

Takes a call recording (video or audio), transcribes it with speaker labels, summarizes it, and writes structured notes into your Obsidian vault.

Requirements

Vault structure — the skill expects these folders inside your Obsidian vault. Folder names are defaults; override them in the Configuration block below if your vault uses different names.

CLI tools — install these before first use, or let Step 0 walk you through it:

Alternative to local: set ELEVENLABS_API_KEY to use ElevenLabs Scribe for transcription + diarization in one call (paid, faster, no setup).

Configuration

The skill reads these variables at runtime. Override any of them via environment variables, or edit the defaults here:

VAULT_ROOT       = $VAULT_ROOT        # auto-detected if not set (see Step 0a)
MEETINGS_DIR     = 03 Meetings
PEOPLE_DIR       = 04 People
DAILY_DIR        = 02 Daily
TEMPLATES_DIR    = _Templates

All paths below are relative to $VAULT_ROOT.

Trigger

When the user provides a call recording (MP4, MOV, WAV, MP3, M4A, etc.) and wants it transcribed, summarized, and documented.

Inputs

• Recording: file path to the recording
• Participants: names of the people on the call
• Date/time: extract from filename if possible, otherwise ask
• Speaker count: default 2, ask if ambiguous

Step 0: Bootstrap check (first run)

Before doing any work, verify the environment is ready. Skip any check that already passes — only prompt the user when something is actually missing.

0a. Resolve the vault root

vault=""
if [ -n "$VAULT_ROOT" ]; then
  vault="$VAULT_ROOT"
else
  dir="$PWD"
  while [ "$dir" != "/" ]; do
    if [ -d "$dir/.obsidian" ]; then vault="$dir"; break; fi
    dir="$(dirname "$dir")"
  done
fi
echo "Vault: ${vault:-NOT FOUND}"

If no vault is found, ask the user:

What's the absolute path to your Obsidian vault? Recommended: use a new, dedicated Obsidian vault for this skill — not your existing personal vault. The skill creates and modifies many notes and folders, and a clean vault avoids polluting your existing notes. If you don't have one yet, create an empty folder, open it in Obsidian (File → Open vault as folder), and paste that path here.

After they answer, validate that <answer>/.obsidian/ exists before using it — if not, warn that the path doesn't look like an Obsidian vault (they may need to open it in Obsidian first) and ask them to confirm or re-enter. Use the validated answer as $VAULT_ROOT for the session (and suggest they set it permanently in their shell profile).

0b. Check required folders

for d in "$MEETINGS_DIR" "$PEOPLE_DIR" "$DAILY_DIR" "$TEMPLATES_DIR"; do
  [ -d "$VAULT_ROOT/$d" ] || echo "MISSING: $d"
done

For each missing folder, ask the user: "Create <folder> in your vault? [y/N]" — if yes, mkdir -p "$VAULT_ROOT/<folder>".

0c. Check required CLI tools

command -v ffmpeg >/dev/null 2>&1 || echo "MISSING: ffmpeg"

If ffmpeg is missing, ask the user before installing. Install command depends on the platform:

• macOS: brew install ffmpeg
• Linux (Debian/Ubuntu): sudo apt install ffmpeg
• Linux (Fedora/RHEL): sudo dnf install ffmpeg
• Windows: download from https://ffmpeg.org/download.html

0d. Install the person template if missing

If $VAULT_ROOT/$TEMPLATES_DIR/new person template.md does not exist, ask the user which version to install:

Install person template — which version?

1. Minimal (default, works in any vault)
2. Full (requires Dataview plugin + Obsidian Bases)

Copy the chosen template from the repo's shared templates/ directory (sibling of this skill dir, i.e. ../templates/) into $VAULT_ROOT/$TEMPLATES_DIR/new person template.md. If the file already exists, leave it alone — the user may have customized it.

Once Step 0 passes, proceed to Step 0.5.

Step 0.5: Determine depth mode

Before transcribing, establish which depth the user wants:

1. Scan the invocation first. If the user's request already specifies a mode, use it and skip the prompt:
   • Words like minimal, fast, quick, --minimal, -m → minimal mode
   • Words like detailed, deep, full, --detailed, -d → detailed mode
2. Otherwise, prompt. No default — if unspecified, ask every time:

Depth?

1. Detailed (best results) — person notes for every person mentioned, including third parties name-dropped mid-call (celebrities, YouTubers, mutual friends). Public figures get researched biographies.
2. Minimal (fast) — person notes for call participants only. Name-drops mid-call stay as dangling wikilinks.

This keeps interactive runs explicit while letting scheduled tasks / cron / /loop pass the mode in the invocation (e.g. /summarize-call ~/call.mp4 minimal) without blocking on input.

The chosen mode determines how Step 6 runs.

Step 1: Choose transcription method

Ask the user:

Transcription method?

1. Local (mlx_whisper + pyannote — free, private, slower, requires setup)
2. ElevenLabs Scribe (paid, faster, handles transcription + diarization in one call)

Option A: Local — walkthrough if not set up

If the user picks local, verify each component and walk them through any missing piece:

1. mlx_whisper

command -v mlx_whisper >/dev/null 2>&1 || echo "MISSING: mlx_whisper"

If missing, ask before installing: pip install mlx-whisper

2. pyannote.audio environment

Check for the venv at ~/.local/share/summarize-call/pyannote-env (persists across reboots, XDG-compliant):

PYANNOTE_ENV="${XDG_DATA_HOME:-$HOME/.local/share}/summarize-call/pyannote-env"
[ -d "$PYANNOTE_ENV" ] || echo "MISSING: pyannote venv"

If missing, walk through setup:

mkdir -p "$(dirname "$PYANNOTE_ENV")"
# Create venv with uv (or python3 -m venv if uv not installed)
uv venv "$PYANNOTE_ENV"
source "$PYANNOTE_ENV/bin/activate"
uv pip install pyannote.audio torch torchaudio

3. HuggingFace token

Check for HF_TOKEN:

[ -n "$HF_TOKEN" ] || echo "MISSING: HF_TOKEN"

If missing, tell the user:

You need a HuggingFace token with access to the pyannote gated repos.

1. Create a token at https://huggingface.co/settings/tokens (choose "Read" scope)
2. Accept the terms for all three repos while logged in:
   • https://huggingface.co/pyannote/speaker-diarization-3.1
   • https://huggingface.co/pyannote/segmentation-3.0
   • https://huggingface.co/pyannote/speaker-diarization-community-1
3. Export for this session: export HF_TOKEN="hf_..."
4. To persist, add that line to your ~/.zshrc (or ~/.bashrc)

Wait for the user to confirm before continuing.

Option B: ElevenLabs Scribe — walkthrough if not set up

If the user picks ElevenLabs, check for the API key:

[ -n "$ELEVENLABS_API_KEY" ] || echo "MISSING: ELEVENLABS_API_KEY"

If missing, tell the user:

You need an ElevenLabs API key.

1. Grab one at https://elevenlabs.io/app/settings/api-keys
2. Export for this session: export ELEVENLABS_API_KEY="..."
3. To persist, add that line to your ~/.zshrc (or ~/.bashrc)

Wait for the user to confirm before continuing.

Before calling Scribe, always print the recording duration and a pricing heads-up so the user can confirm:

This recording is <HH:MM:SS> (<minutes> min). Check ElevenLabs pricing at https://elevenlabs.io/pricing for the current per-minute rate on the Scribe model. Continue? [y/N]