Obsidian Vault Writer — Cross-Agent Integration Handoff

[home-path]
├── __init__.py          # Package init, version 0.1.0
├── config.py            # VAULT_PATH, directory names, filter lists
├── slugify.py           # Unicode title → filesystem-safe slug
├── templates.py         # Markdown templates for 8 note types
├── linker.py            # [[wikilink]] extraction from synthesis JSON
├── writer.py            # VaultWriter class + CLI entry point (4 modes)
├── backfill.py          # One-time kimi_memory.db migration
├── requirements.txt     # pyyaml>=6.0
├── ARCHITECTURE.md      # Full architecture documentation
└── INTEGRATION-HANDOFF.md  # This file

[home-path]
├── Inbox/          # Real-time synthesis notes, by date
│   └── YYYY-MM-DD/
│       └── {HHMMSS}-{slug}.md
├── Projects/       # One note per project (evergreen index)
├── Entities/       # One note per knowledge graph entity
├── Concepts/       # Dream seed ideas
├── Sessions/       # Agent session summaries (ALL agent types)
├── Daily/          # Auto-generated daily rollups
├── Knowledge/      # Curated: Facts.md, Preferences.md, Patterns.md
└── Templates/      # 5 Obsidian templates

echo '{"agent_type":"codex","session_id":"xyz-123","provider":"openai","model":"o3-mini","goal":"Fix auth bug","outcome":"complete","date":"2026-03-01","stats":{"duration_minutes":12.5,"prompt_count":5,"tool_call_count":30,"files_modified":["auth.py","config.py"],"files_read":["models.py"],"aao_quality":"high","aao_confidence":0.9,"aao_device_score":0.88},"project_refs":["comp-core"],"key_topics":["authentication"],"source":"codex-daemon"}' \
  | python3 -m obsidian_vault_writer.writer --mode agent-session

import sys
sys.path.insert(0, str(Path.home() / "projects"))

from obsidian_vault_writer.writer import VaultWriter

writer = VaultWriter()  # Uses VAULT_PATH env or [home-path]

# Agent session note
path = writer.write_agent_session({
    "agent_type": "gemini",        # claude-code | codex | gemini | clawdbot | human | custom
    "session_id": "abc-123",
    "provider": "google",           # anthropic | openai | google | together
    "model": "gemini-2.5-pro",
    "goal": "Research N'Ko morphology",
    "outcome": "complete",           # complete | failed | cancelled | interrupted | degraded
    "date": "2026-03-01",
    "stats": {
        "duration_minutes": 25.0,
        "prompt_count": 8,
        "tool_call_count": 15,
        "files_modified": ["research.md"],
        "files_read": ["corpus.txt"],
        # Optional AAO fields (if task was AAO-gated):
        "aao_quality": "high",       # high | medium | low | failed
        "aao_confidence": 0.85,
        "aao_device_score": 0.91,
    },
    "project_refs": ["nko-linguistics"],  # Creates [[project]] links + Project stubs
    "key_topics": ["morphology", "NKo"],  # Creates [[topic]] links
    "source": "gemini-daemon",
})
print(f"Written: {path}")
# → [home-path]

path = writer.write_synthesis(
    synthesis={
        "enriched_prompt": "Full context-enriched prompt...",
        "intent": "idea",
        "confidence": 0.85,
        "route": "direct",
        "dream_seeds": [{"title": "Voice NKo Keyboard", "energy": 0.7, "tags": ["NKo"]}],
        "skill_chain": ["lin:nko", "thk:quantum"],
        "knowledge_connections": [
            {"subject": "NKo keyboard", "predicate": "related_to", "object": "voice recognition"}
        ],
        "learnings": {"facts": [{"key": "nko_voice", "value": "requires tone-aware recognition"}]},
        "project_refs": ["nko-linguistics"],
    },
    message="Original user message text",
    channel="discord:ideas",
)

{
  "agent_type": "claude-code",
  "session_id": "unique-session-id",
  "provider": "anthropic",
  "model": "claude-opus-4-6",
  "goal": "What was being worked on (first 120 chars)",
  "outcome": "complete",
  "date": "2026-03-01",
  "stats": {
    "duration_minutes": 45.3,
    "prompt_count": 12,
    "tool_call_count": 87,
    "files_modified": ["file1.py", "file2.rs"],
    "files_read": ["config.py"],
    "estimated_tokens": 15000,
    "aao_quality": "high",
    "aao_confidence": 0.85,
    "aao_device_score": 0.912
  },
  "project_refs": ["comp-core", "nko-linguistics"],
  "key_topics": ["authentication", "graph-kernel"],
  "source": "session_end_hook"
}

---
type: agent-session
agent_type: claude-code
provider: anthropic
model: claude-opus-4-6
session_id: abc123def456
date: 2026-03-01
outcome: complete
source: session_end_hook
duration_minutes: 45.3
prompt_count: 12
tool_calls: 87
aao_quality: high
aao_confidence: 0.85
aao_device_score: 0.912
project_refs: [comp-core]
tags: [authentication, graph-kernel]
created: 2026-03-01T18:30:00Z
---

# Claude Code Session: Fix authentication bug in login flow

| | |
|---|---|
| **Agent** | Claude Code (claude-opus-4-6) |
| **Duration** | 45m |
| **Prompts** | 12 |
| **Tool Calls** | 87 |
| **Outcome** | complete |
| **AAO Quality** | high (85% confidence) |
| **Device Score** | 0.91 |

## Goal
Fix authentication bug in login flow

## Files Modified (2)
- `auth.py`
- `config.py`

## Files Read (1)
- `models.py`

## Links
- Project: [[comp-core]]
- Related: [[authentication]]
- Related: [[graph-kernel]]

---
type: synthesis
intent: idea
confidence: 0.85
route: direct
channel: discord
project_refs: [nko-linguistics]
tags: [NKo, voice-recognition]
created: 2026-03-01T10:30:00Z
---

# Voice-Controlled NKo Keyboard

## Enriched Prompt
Context-rich synthesis...

## Dream Seeds
- **Voice-Controlled NKo Keyboard** (energy: 0.7)
  Tags: NKo, voice recognition

## Knowledge Connections
- [[NKo keyboard]] --related to--> [[voice recognition]]

## Links
- Project: [[nko-linguistics]]
- Related: [[NKo keyboard]], [[voice recognition]]

Discord/SMS/Telegram task
    ↓
mac_tasks row created in Supabase
    ↓
AAO Ticket Gate (mints HMAC token, every 3min)
    ↓
AAO Dedup Filter (fingerprints, blocks duplicates, every 5min)
    ↓
pulse-dispatcher (claims + routes to device)
    ↓
Agent executes on device (Mac1-5, cloud-vm)
    ↓
Task completes → mac_tasks.status = 'complete'
    ↓
AAO Reputation Collector (every 10min)
    ├── Assess quality (output length, exit code, errors, duration)
    ├── Sign HMAC attestation → Graph Kernel
    ├── Update device reputation score
    └── write_task_to_vault() ← NEW
         ├── Map claimed_by → agent_type
         ├── Infer provider from model_used
         ├── Extract project_refs from project_path
         ├── Include aao_quality, aao_confidence, aao_device_score
         └── Sessions/{date}-{agenttype}-{task_uuid}.md
              ↓
         obsidian-sync → all devices

SELECT id, task_content, output, exit_code, duration_ms, error_log,
       claimed_by, completed_at, started_at, task_type, model_used,
       project_path, source, session_id, pool_mode, team_role
FROM mac_tasks
WHERE status = 'complete' AND completed_at > $last_assessed_at
ORDER BY completed_at DESC LIMIT 30

# Priority 1: Model string override
if "gpt" or "o1" or "o3" in model_used → agent_type = "codex"
if "gemini" in model_used             → agent_type = "gemini"

# Priority 2: Device mapping
claimed_by = "mac1" through "mac5"    → agent_type = "claude-code"
claimed_by = "cloud-vm"              → agent_type = "claude-code"
claimed_by = "codex"                 → agent_type = "codex"

# Fallback
else                                  → agent_type = "custom"

# At top of flow file
import sys
from pathlib import Path
VAULT_WRITER_DIR = Path.home() / "projects"
if str(VAULT_WRITER_DIR) not in sys.path:
    sys.path.insert(0, str(VAULT_WRITER_DIR))

# In your task function
from obsidian_vault_writer.writer import VaultWriter
writer = VaultWriter()
path = writer.write_agent_session({...})  # or write_synthesis(), write_daily_summary()

                    Source Systems
    ┌─────────────────────────────────────────────┐
    │  Discord    Claude Code   AAO Tasks   Prefect│
    │  Synthesis  SessionEnd    Reputation  Flows  │
    └──────┬──────────┬───────────┬──────────┬─────┘
           │          │           │          │
           ▼          ▼           ▼          ▼
    ┌──────────────────────────────────────────────┐
    │     obsidian_vault_writer (Python module)     │
    │                                               │
    │  write_synthesis()  write_agent_session()      │
    │  write_daily_summary()  ensure_entity()        │
    │  ensure_project()  ensure_concept()            │
    └───────────────────┬──────────────────────────┘
                        │ writes .md files
                        ▼
    ┌──────────────────────────────────────────────┐
    │     [home-path] (cloud-vm filesystem)   │
    │                                               │
    │  Syncthing ←→ Mac1                            │
    │  ob sync --continuous → Obsidian Cloud        │
    └───────────────────┬──────────────────────────┘
                        │ Obsidian Sync (E2E)
           ┌────────────┼────────────┐
           ▼            ▼            ▼
        iPhone       Mac4/Mac5    Any device
        Obsidian     Obsidian     with Obsidian

# Pipe your session data to the vault writer
echo '{"agent_type":"YOUR_TYPE","session_id":"YOUR_SESSION","provider":"YOUR_PROVIDER","model":"YOUR_MODEL","goal":"What you worked on","outcome":"complete","date":"'$(date +%Y-%m-%d)'","stats":{"duration_minutes":10,"prompt_count":5,"tool_call_count":20,"files_modified":["file.py"],"files_read":["config.py"]},"project_refs":["project-name"],"key_topics":["topic"],"source":"YOUR_SOURCE"}' \
  | python3 -m obsidian_vault_writer.writer --mode agent-session