Demo with LLM Wiki

1.txt

# ติดตั้ง และ ทำการ setup ผ่าน wizard แบบ step-by-step
$pip install obsidian-llm-wiki
$olw setup

╭────────────────────────────────────────────────────╮
│    obsidian-llm-wiki v0.2.0  ·  first run setup    │
╰────────────────────────────────────────────────────╯

  Step 1/4  Ollama connection
    Warning: Could not reach http://localhost:11434
    You can still configure manually — run olw doctor later.
    Ollama URL (http://localhost:11434): 

  Step 2/4  Fast model  (analysis & routing · 3–8B recommended)
    (e.g. gemma4:e4b, llama3.2:3b, qwen2.5:14b)
    Model name (gemma4:e4b): 

  Step 3/4  Heavy model  (article writing · 7–14B recommended)
    (e.g. gemma4:e4b, llama3.2:3b, qwen2.5:14b)
    Model name (qwen2.5:14b): 

  Step 4/4  Default vault path  (press Enter to skip)
    Vault path (): ~/mywiki

╭──────────────────────────────────────────────────────╮
│  ✓  Setup complete                                   │
│                                                      │
│    Fast model:   gemma4:e4b                          │
│    Heavy model:  qwen2.5:14b                         │
│    Ollama:       http://localhost:11434              │
│    Vault:        ~/mywiki                             │
│                                                      │
│    Next steps:                                       │
│      olw init /mywiki                                │
│      olw run  (or: olw ingest --all && olw compile)  │
╰──────────────────────────────────────────────────────╯

2.txt

# ทำการ initial project สำหรับจัดการ wiki ขึ้นมา
$olw init <path-to-llm-wiki>

Created fresh vault structure
INFO     Initialised git repo at
Vault initialised: ~/llm-wiki
Next steps:
  1. Drop .md notes into raw/
  2. Run olw run  (ingest + compile + lint in one step)
  3. Review drafts: olw review

# โครงสร้างของ project เป็นดังนี้
├── raw
├── vault-schema.md
├── wiki
│   ├── INDEX.md
│   └── sources
└── wiki.toml

3.txt

# ทำการ ingest ข้อมูลจาก /raw folder เพื่อทำการแปลงไปยัง folder /wiki
$olw ingest raw/{{filename}}.md

INFO     Note Harness engineering for coding agent users.md split into 3 chunks   
         for analysis (17221 chars, chunk_size=8192)                              
INFO     Analyzing Harness engineering for coding agent users.md [part 1/3] …     
INFO     Analyzed Harness engineering for coding agent users.md [part 1/3] (10.1s)
INFO     Analyzing Harness engineering for coding agent users.md [part 2/3] …     
INFO     Analyzed Harness engineering for coding agent users.md [part 2/3] (3.8s) 
INFO     Analyzing Harness engineering for coding agent users.md [part 3/3] …     
INFO     Analyzed Harness engineering for coding agent users.md [part 3/3] (2.2s) 
INFO     Source summary written: Harness Engineering For Coding Agent Users.md    
INFO     Ingested: Harness engineering for coding agent users.md (quality=medium, 
         concepts=['Harness definition (Agent = Model + Harness)', 'Outer harness 
         goals (increase correctness, provide feedback loop, reduce toil)',       
         'Computational vs Inferential guidance/sensors'])                        
  Harness engineering for coding agent users.md ━━━━━━━━━━━━━━━━━━━━━━ 1/1 0:00:16
Done. Ingested: 1  Skipped: 0  Failed: 0

# Watch mode ถ้ามีการเปลี่ยนแปลง จะทำการแปลงให้ทันที
$olw watch

# Run all processes
$olw run

4.txt

wiki
├── INDEX.md
├── log.md
└── sources
    └── Harness Engineering For Coding Agent Users.md

2 directories, 3 files

5.txt

$olw query "what is harness engineer"

Sources: Harness Engineering For Coding Agent Users

A harness refers to everything surrounding an AI agent, excluding the model       
itself, which can be tailored to specific contexts like coding agents. A          
well-built outer harness aims to increase correctness, provide self-correction    
feedback loops, and reduce human review toil. It integrates both computational    
(fast, deterministic) and inferential (slower, semantic) guidance and feedback    
mechanisms, steered by a human. Key concepts related to this include [[Harness    
definition (Agent = Model + Harness)]], [[Outer harness goals (increase           
correctness, provide feedback loop, reduce toil)]], [[Computational vs Inferential
guidance/sensors]], [[Feedforward vs Feedback controls]], [[Steering loop (human  
iteration)]], [[Change lifecycle timing (shift-left principle)]], [[Continuous    
drift and health sensors]], and [[Regulation categories (Maintainability          
harness)]].

demo.txt

---
source_title: "Harness engineering for coding agent users"
source_url: "https://martinfowler.com/articles/harness-engineering.html"
captured: "2026-04-15T22:13:05+07:00"
---
The term harness has emerged as a shorthand to mean everything in an AI agent except the model itself - [Agent = Model + Harness](https://blog.langchain.com/the-anatomy-of-an-agent-harness/). That is a very wide definition, and therefore worth narrowing down for common categories of agents. I want to take the liberty here of defining its meaning in the bounded context of using a coding agent. In coding agents, part of the harness is already built in (e.g. via the system prompt, or the chosen code retrieval mechanism, or even a [sophisticated orchestration system](https://www.anthropic.com/engineering/effective-harnesses-for-long-running-agents)). But coding agents also provide us, their users, with many features to build an outer harness specifically for our use case and system.

wiki.toml

[models]
fast = "gemma4:e4b"
heavy = "qwen2.5:14b"
# Optional: set heavy = fast to use a single model for everything

[ollama]
url = "http://localhost:11434"
timeout = 600
fast_ctx = 16384                  # context window for fast model (tokens)
heavy_ctx = 32768                 # context window for heavy model (tokens)

[pipeline]
auto_approve = false
auto_commit = true
auto_maintain = false
watch_debounce = 3.0
max_concepts_per_source = 8
ingest_parallel = false   # true = parallel chunks (needs OLLAMA_NUM_PARALLEL>=4)