← The Independent 100: a study

geoff (Geoffrey Huntley)

#90 in The Builders on Main, the curator’s rank and section @GeoffreyHuntley Relevance to our work: 5 of 5

Bracketed tags are evidence labels: [Verified] means we read the page, file or API response ourselves, [Reported] means a source says so and we did not check the underlying thing, and [Inference] is our own reasoning.

Credit

The Independent 100 list was curated by dan (@irl_danB) and published on 2026-09-24 at independent.prose.md (announcement). The ranks and section names are his, and number 100 is his open wild-card slot. This study is ours and is not endorsed by him or by anyone on the list.

What they publicly do

He works at Antithesis on AI and verifying outcomes [Verified: ghuntley.com/slop], and defined the Ralph loop [Verified: ghuntley.com/ralph]. Ralph in its simplest form is 'while :; do cat PROMPT.md | claude-code ; done' [Verified: ghuntley.com/ralph]. Publishes a coding-agent workshop and new MIT Rust proxies [Verified: GitHub].

Public work

  • Ralph Wiggum as a 'software engineer' (the Ralph loop) blog no license stated last activity 2025-07-14 (original); 2026-03-15 (porting)

    Defines Ralph: a bash loop feeding the same PROMPT.md to a coding agent until the work converges; follow-ups 'everything is a ralph loop' (2026-01-17) and a porting recipe (2026-03-15) that turns tests and sources into cited /specs via subagents, then loops on the port.

  • how-to-build-a-coding-agent course no license file last activity 2026-09-20 · 5,845 GitHub stars on 2026-09-24

    Go workshop that builds a coding agent step by step (tools, loop, file edits).

  • preflight repo MIT last activity 2026-09-20 · 76 GitHub stars on 2026-09-24

    Rust local proxy between the harness and the model: decodes requests and attachments (images, PDFs, OCR), replaces detected credentials with [REDACTED:rule-id] placeholders, blocks when inspection is impossible; rule-based, not entropy-based.

  • underclass repo MIT last activity 2026-09-23 · 165 GitHub stars on 2026-09-24

    Rust OpenAI-compatible pooling proxy: pins a session to one account to keep the prompt cache warm, cools quota-exhausted subscriptions until reset, fails fast with the earliest Retry-After.

  • cursed repo MIT last activity 2025-11-16; v0.0.1 released 2025-09-09 · 653 GitHub stars on 2026-09-24

    A Gen-Z-slang programming language built by running Claude in a loop for three months (ghuntley.com/cursed).

  • loom repo no license file last activity 2026-04-10 · 1,376 GitHub stars on 2026-09-24

    Rust coding agent workspace (30+ crates); README says do not use unless you are the author.

Licenses of GitHub repositories are from the authenticated GitHub license API on 2026-09-25, one row per repository in evidence/licenses.jsonl (“no license file” means the API found none; “license file present, not recognised by GitHub” means GitHub could not classify it). Other licenses are as the source states. Check the license file before reusing anything.

Relevance to our work

5 of 5 lenses: agents, orchestration, verification, security, research-method, writing

The Ralph loop is the root of much agent loop tooling, including Ian Nuttall's ralph templates. preflight (MIT, Rust) is an adoptable outbound secret-redaction layer. His porting recipe is directly comparable to the port methods franken-research grades, and his current posts push property-based and deterministic simulation testing.

Deep dive:

Scored 0 to 5 against the study’s relevance rubric: 5 means public, openly licensed work squarely in our lanes that we could study or adopt now, and 1 means little connection to our work. The score is one reviewer’s judgement of how much this public work bears on our own work. It is not a ranking of people and says nothing about the person or the quality of their work.

What we would study next

  1. The porting recipe: tests to cited specs via subagents, then sources to specs, then a Ralph loop on the port (ghuntley.com/porting). A short, public method to set beside Jeffrey's four-document port method in franken-research packets (spec citations, oracle, convergence). Effort: small.
  2. preflight as an outbound secret filter in front of coding harnesses (github.com/ghuntley/preflight). Adds a request-side guard (including OCR of screenshots and PDFs). Created 2026-09-20: test it before trusting it. Effort: medium.
  3. how-to-build-a-coding-agent workshop (github.com/ghuntley/how-to-build-a-coding-agent). A minimal reference agent loop for onboarding and training content; no license, so link to it rather than copy. Effort: small.

Before reusing anything: preflight, underclass and cursed are MIT [Verified: GitHub license API, 2026-09-25, evidence/licenses.jsonl]. loom is marked 'SPDX-License-Identifier: Proprietary' and 'do not use'. how-to-build-a-coding-agent and sup have no license file (study only). noyaml and groundhog are AGPL-3.0. underclass pools subscription accounts; check each provider's terms before any use [Inference]. The opening of the 'back pressure' post (ghuntley.com/pressure) is public and credits Moss's essay on back pressure; the rest of the post is for subscribers [Verified, read 2026-09-25]. The how-to-ralph-wiggum repository under his account is a fork of Clayton Farr's ralph-playbook (github.com/ClaytonFarr/ralph-playbook), so that playbook is Clayton Farr's work.

Sources

Overall confidence in this record: High. Sources were read on 2026-09-24 and every link was checked on 2026-09-25. The full record is line 90 of people.jsonl.

Limits

Not verified: Whether preflight and underclass work as described (both created 2026-09-20; not run). Claims that Ralph can replace most outsourcing for greenfield work. The subscriber-only parts of the 'back pressure' post and of the 'specs' post (ghuntley.com/specs). Bombadil (antithesishq, MIT) and Hegel, which he promotes in recent posts, are Antithesis/hegeldev projects, not his own; only his hegel-lean and hegel-erlang SDKs (MIT) are his.

This record comes from a first-pass survey of public material by an AI agent session on 2026-09-24, edited for publication on 2026-09-25. Nothing was installed or run for it. Roles, stars and licenses change; see the study’s limits.

Request a correction

Is something about this record wrong or out of date, or would you rather not be covered? Open a correction (a GitHub issue form with this page’s address filled in). A request to remove a record is honoured without argument; accepted corrections are recorded in the changelog with the date.

Help fix it

Each opens a GitHub issue form with this page’s address filled in. You need a GitHub account.

Made by Joshua Nowak (ZestStream) with AI coding agents. Not affiliated with or funded by Jeffrey Emanuel; Joshua uses his public tools and is a paying subscriber to jeffreys-skills.md.

Source on GitHub MIT licensed