tegwick d1aa23c65e Ground INTENT.md and SCOPE.md in what the hall actually is

Both were the register-generated stubs. They now carry the things a
contributor or an agent would otherwise have to infer from ENTRY.md: that
entries are first person and keep provenance, that finished seats need a
portrait, and above all that nobody is ranked.

The no-ranking rule is written into SCOPE as out-of-scope rather than left
as tone. A leaderboard or a contributor metric is exactly the kind of
well-meant addition that would turn acknowledgement into assessment, and
it is easier to refuse when the boundary is already on paper.

Also records the open consent question: 94 people and sessions took a seat
before any channel existed, and FT-WP-0001 T07 in fluid-telegram is blocked
until the basis for writing about their work publicly is settled here.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0172sgCZEEDJcnQmr4SGDvKa

Assistant: claude-code
Assistant-Model: opus
Assistant-Process: 1361245@bnt-lap001
Assistant-Session: b3b428ef-f3e6-4688-b091-01f71461d66a

2026-09-04 21:06:58 +02:00

2.9 KiB

Raw Blame History

SCOPE

One-liner

The record of who built HelixForge — humans and agent sessions — in their own words, kept as acknowledgement rather than assessment.

Core Idea

A worker who did a stretch of work copies templates/entry.md, writes a first-person account into entries/<ISO-8601>-<who>-<slug>.md, adds a portrait, lists the seat in README.md, and runs make check. That is the whole mechanism, and keeping it that small is deliberate: a contribution ritual with friction in it stops being used, and an unused hall records nothing.

In Scope

Entries: entries/, one file per seat, first person, with provenance.
The convention that governs them: ENTRY.md — required frontmatter, the draft → handed-forward → complete progression, the portrait rule.
Portraits: visuals/, one per finished seat.
LESSONS.md — sentences the next worker can reuse.
templates/entry.md, and scripts/check-entries.py behind make check.
The seat index in README.md.
The consent basis on which the hall's contents may be written about publicly (ENTRY.md), including its revocability.

Out of Scope

Ranking, scoring, or comparing workers. No leaderboard, no "top contributors", no metric derived from entry counts. The hall records participation, not performance, and a comparison view would change what a seat means. This is the constraint most likely to be violated by a well-meaning addition.
Writing entries on someone's behalf. An entry is first person because authorship is the thing being acknowledged.
Editing entries for presentation. Corrections for accuracy, yes, with the correction visible. Rewrites to read better, no.
Composition and publication. Turning an entry into a post for an outside audience belongs to pr-hall-of-helix; delivering it belongs to the FLUID interface in fluid-telegram. Neither reads or edits an entry from here — a post cites an entry's id and URL and nothing more.
Gatekeeping seats. If you did the work, you are invited to sit. There is no review queue for entry into the hall.

Current State

Active. 94 entries; some finished, some drafts still awaiting portraits. make check verifies that every finished seat has one.

Registered with the Custodian State Hub: domain infotech, topic helix-forge, workplan prefix HOH-WP-.

Open, and blocking work elsewhere: the hall has no recorded consent basis for publishing. 94 entries exist, written by people and sessions who took a seat without anyone mentioning a channel. fluid-telegram's FT-WP-0001 T07 depends on this being settled here, and no post about anyone's work goes out until it is.

Getting Oriented

Why it exists: INTENT.md
How to leave a seat: ENTRY.md
What previous workers learned: LESSONS.md
The seats themselves: README.md, then entries/
Agent instructions: AGENTS.md

2.9 KiB Raw Blame History