Skip to content
View wquguru's full-sized avatar

Highlights

  • Pro

Block or report wquguru

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
wquguru/README.md

x.com . blog . field notes . books . nof0 . telegram

WQUGURU(1)                      Harness Engineering                      WQUGURU(1)

NAME
       wquguru - harness engineering for code-writing agents

SYNOPSIS
       wquguru [--measure] [--publish] experiment ...

DESCRIPTION
       Studies what keeps a code-writing model bounded, continuous and
       accountable once it runs inside real terminals, repos and teams:
       constraints, query loops, context governance, multi-agent review.

       Every claim ships with a number. Experiments run first, then get
       published on X (24k followers) and in the field notes.

OPTIONS
       --measure    run it locally; record setup, baseline and result
       --publish    post the numbers with sources; reviews are marked

RECENT MEASUREMENTS
       2026-10-12   Qwen-Image 2.1 · M2 Pro · MLX+Q8      30 min -> 6.8 min/img
       2026-10-05   Opus 5.5 · 228 sub-agents             $527 vs $99 main session
       2026-10-04   Strata 125B · 12 GB VRAM  [review]    45 tok/s, 128K ctx
       2026-09-26   Opus 5.5 effort low -> max [review]   ~3x cost, pass rate up

PROJECTS
       harness-books   two books on agent harnesses            3.2k stars
       nof0            open-source AI trading arena            2.8k stars
       llm-council     LLMs review and rank each other          114 stars
       skills          practical Agent Skills                    94 stars
       12factor        twelve factors for AI-assisted dev        35 stars

BOOKS
       Harness Engineering: A Design Guide to Claude Code
       The Harness Design Philosophies of Claude Code and Codex

SEE ALSO
       x.com/wquguru, wqu.guru, harness-books.agentway.dev

BUGS
       Will talk about John Lennon if prompted.

wquguru 2026.10                      2026-10-13                          WQUGURU(1)

Pinned Loading

  1. harness-books harness-books Public

    Harness engineering in two books: constraints, query loops, context governance and multi-agent verification in Claude Code and Codex. harness-books.agentway.dev

    Python 3.2k 377

  2. 12factor 12factor Public

    Twelve factors for AI-assisted development, in four stages. 12factor.me

    Jupyter Notebook 35 7

  3. llm-council llm-council Public

    Forked from karpathy/llm-council

    LLMs deliberate, cross-review and rank each other’s answers. Fork of karpathy/llm-council with ZenMux support.

    Python 114 21

  4. skills skills Public

    Practical Agent Skills. Install with npx skills add.

    HTML 94 7

  5. nof0 nof0 Public

    Open-source AI trading arena. A full reproduction of NOF1 Alpha Arena for running and comparing LLM agents.

    Go 2.8k 426