meeting-ingestion

Files (29)

• SKILL.md
• me.yaml.template · config/
• orgs.yaml.template · config/
• intake-v1.schema.json · references/
• manifest-v3.schema.json · references/
• meeting-id-convention.md · references/
• quality-gate-docs.md · references/
• archive.py · scripts/
• calendar_match.py · scripts/
• classifier.py · scripts/
• config_loader.py · scripts/
• identity.py · scripts/
• ingest.py · scripts/
• intake_checker.py · scripts/
• meeting_auto_process.py · scripts/
• meeting_cli.py · scripts/
• memory_router.py · scripts/
• normalize_inbox.py · scripts/
• org_classifier.py · scripts/
• paths.py · scripts/
• pocket_adapter.py · scripts/
• process.py · scripts/
• pull.py · scripts/
• quality_gate.py · scripts/
• router.py · scripts/
• run_shapes.py · scripts/
• title_normalizer.py · scripts/
• validate_manifest.py · scripts/
• zth_scanner.py · scripts/

SKILL.md (first lines)

# Meeting Ingestion Skill v3

Complete meeting transcript processing pipeline with multi-source intake: Google Drive, Fathom, Fireflies, Pocket, and manual transcript import. Unified CLI for transcript processing from ingest to archive, plus intake health checks and historical backfill.

## Canonical Processing Model

The current canonical generation surface is **S-shape based**.

- `file 'Skills/meeting-ingestion/scripts/run_shapes.py'` is the primary shape generator
- `file 'N5/lib/meeting_shapes.py'` defines shape selection, filenames, prompts, and post-processors
- `file 'Skills/meeting-ingestion/scripts/process.py'` as the orchestration bridge that routes gated meetings into canonical shape generation by default

Legacy B-block scripts remain in the repo only for backward compatibility, manual override flows, and migration support. They are not the canonical generation model.

## Quick Start

```bash
# Full pipeline for new transcript
python3 Skills/meeting-ingestion/scripts/meeting_cli.py ingest /path/to/transcript.md
python3 Skills/meeting-ingestion/scripts/meeting_cli.py identify <meeting-folder>
python3 Skills/meeting-ingestion/scripts/meeting_cli.py gate <meeting-folder>
python3 Skills/meeting-ingestion/scripts/meeting_cli.py process <meeting-folder>
python3 Skills/meeting-ingestion/scripts/meeting_cli.py archive --execute

# Or use orchestrator (processes entire pipeline)
python3 Skills/meeting-ingestion/scripts/meeting_cli.py tick

# Intake management
python3 Skills/meeting-ingestion/scripts/meeting_cli.py check
python3 Skills/meeting-ingestion/scripts/meeting_cli.py backfill --source all --from 2026-02-01 --to 2026-03-10 --dry-run
python3 Skills/meeting-ingestion/scripts/meeting_cli.py drive-import --scan
```

## v3 Pipeline Overview

```
┌─────────────────────┐
│ Google Drive        │
│   ├─ transcript.md  │
│   └─ meeting.jsonl  │
└─────────┬───────────┘
          │ pull
          ▼
┌─────────────────────┐
│ RAW FILES           │
│ (Inbox root)        │
└─────────┬───────────┘
          │ ingest
          ▼
┌─────────────────────────────────────────┐
│ INGESTED                                │
│  YYYY-MM-DD_Participant/                │
│    ├─ transcript.md                     │
│    └─ manifest.json {status: ingested}  │
└─────────┬───────────────────────────────┘
          │ identify
          ▼
┌─────────────────────────────────────────┐
│ IDENTIFIED                              │
│  YYYY-MM-DD_Participant/                │
│    ├─ transcript.md                     │
│    └─ manifest.json {status: identified}│
│        ├─ participants: [...]           │
│        ├─ meeting_type: external        │
│        └─ calendar_match: {...}         │
└─────────┬───────────────────────────────┘
          │ gate
          ▼
┌─────────────────────────────────────────┐
│ GATED                                   │
│  YYYY-MM-DD_Participant/                │
│    ├─ transcript.md                     │
│    └─ manifest.json {status: gated}     │
│        └─ quality_gate: {passed: true}  │
└─────────┬───────────────────────────────┘
          │ process
          ▼
┌─────────────────────────────────────────┐
│ PROCESSED                               │
│  YYYY-MM-DD_Participant/                │
│    ├─ transcript.md                     │
│    ├─ S01_CONTEXT.md                    │
│    ├─ S02_ACTIONS.md                    │
│    ├─ S03_EVIDENCE.md                   │
│    ├─ S04_PEOPLE.md                     │
│    ├─ S05_WISDOM.md                     │
│    ├─ S06_RECAP.md                      │
│    ├─ S07_ZO_DIRECTIVES.md              │
│    ├─ S08_ZO_INTEL.md                   │
│    ├─ PP_GRAPH_EDGES.jsonl              │
│    └─ manifest.json {status: processed} │
└─────────┬───────────────────────────────┘
          │ closeout + archive
          ▼
┌─────────────────────────────────────────┐
│ ARCHIVED                                │
│  Week-of-YYYY-MM-DD/external/           │
│    └─ YYYY-MM-DD_Participant/           │
└─────────────────────────────────────────┘
```

## Commands Reference

### v3 Pipeline Commands (Recommended)

#### `ingest` - Raw → Ingested
Convert raw transcripts into meeting folders with v3 manifest.

```bash
python3 meeting_cli.py ingest <path> [--dry-run] [--json]

# Examples
python3 meeting_cli.py ingest transcript.md          # Single file 
python3 meeting_cli.py ingest meeting-folder/       # Folder with transcript
python3 meeting_cli.py ingest . --dry-run          # Preview all in current dir
```

**What it does:**
- Detects format (JSONL vs markdown vs folder)

From thevibethinker/n5os-ode, MIT.

More from thevibethinker/n5os-ode