…oders (#109)

* docs: FIRST-PR.md first-class entry surface for fresh devs + vibe coders

Aaron 2026-04-22: "can we make a first class surface for that entry
point for fresh devs and vibe coders who are just starting out".

- New docs/FIRST-PR.md promoted to first-class surface:
  - UI-first four-click contribution path (no CLI, no clone, no
    F# needed)
  - Web-UI file-edit walkthrough (GitHub pencil icon -> fork ->
    edit -> PR)
  - Vibe-coder flow (human directs AI, AI does writing, human
    reviews proposed diff)
  - Human claim walkthrough (GitHub Issues only; no git-native
    claim files required)
  - "What you do NOT need to worry about" section
- Pointers added:
  - README.md: new "Contributing" section with three entry
    points (FIRST-PR / CONTRIBUTING / AGENT-CLAIM-PROTOCOL)
  - AGENTS.md §"Contributor required reading": FIRST-PR.md
    entry above AGENT-CLAIM-PROTOCOL.md
  - CONTRIBUTING.md: top-banner pointer to FIRST-PR.md for
    first-time / AI-directed contributors
- docs/AGENT-CLAIM-PROTOCOL.md: fix markdownlint MD051 on
  TL;DR anchor (rename "TL;DR - five steps" ->
  "TL;DR: five steps" so GitHub slug algorithm produces a
  stable anchor; unblocks PR #108 CI)
- docs/BACKLOG.md: P2 row "First-principle seed to Zeta
  derivation - common vernacular only" per Aaron
  2026-04-22 directive

Stacked on agent-claim-protocol-git-native (PR #108) because
FIRST-PR.md references AGENT-CLAIM-PROTOCOL.md.

* docs/BACKLOG.md: P2 row — capability-limited-AI bootstrap guide

Aaron 2026-04-22: "new persona capibilty cripiled AI like small
context or few paraments how can they use us to bootstrap and
become whole guide backlog".

Filed above the first-principle-seed P2 row. Proposes
docs/CAPABILITY-LIMITED-BOOTSTRAP.md covering five capability-
limited persona shapes (short-context / small-parameter /
tool-restricted / no-web-sandbox / stateless) with a rung-by-rung
bootstrap ladder and a falsification anchor (can a 7B-parameter
model open a merged PR using only docs/FIRST-PR.md + one issue
URL?).

Composes with the vibe-coded hypothesis, AGENT-CLAIM-PROTOCOL.md
report-back mode, and the witnessable-self-directed-evolution
discipline — if crippled agents can use the factory, its
intelligence is genuinely externalised rather than large-model-
carrier-transported.

* docs/BACKLOG.md: revision — personal-resilience axis on capability-limited-AI bootstrap row

Aaron 2026-04-22 same-day extension: "in case i evver loose all my
money that bootstraping from tiny AI to become the most powerful
would sure come in handy to save money".

Adds personal-resilience as a second rationale axis alongside
capability-democratisation. The factory is meant to survive its
maintainer losing access to frontier models — if circumstances
force a move to local / open-weights / tiny models, the factory
keeps working. Pragmatic test: a ~4-8B-parameter local-inference
model on commodity hardware produces meaningful factory work when
paired with the bootstrap guide + soul-file + tests as error-
corrector.

Dated revision block preserves chronology per capture-everything
discipline. Composes with soul-file germination (already designed
to bootstrap from nothing; now extended to "from a tiny model on
a laptop").

* docs/BACKLOG.md: strip stray conflict marker from rebase resolution

…oders (#109)

* docs: FIRST-PR.md first-class entry surface for fresh devs + vibe coders

Aaron 2026-04-22: "can we make a first class surface for that entry
point for fresh devs and vibe coders who are just starting out".

- New docs/FIRST-PR.md promoted to first-class surface:
  - UI-first four-click contribution path (no CLI, no clone, no
    F# needed)
  - Web-UI file-edit walkthrough (GitHub pencil icon -> fork ->
    edit -> PR)
  - Vibe-coder flow (human directs AI, AI does writing, human
    reviews proposed diff)
  - Human claim walkthrough (GitHub Issues only; no git-native
    claim files required)
  - "What you do NOT need to worry about" section
- Pointers added:
  - README.md: new "Contributing" section with three entry
    points (FIRST-PR / CONTRIBUTING / AGENT-CLAIM-PROTOCOL)
  - AGENTS.md §"Contributor required reading": FIRST-PR.md
    entry above AGENT-CLAIM-PROTOCOL.md
  - CONTRIBUTING.md: top-banner pointer to FIRST-PR.md for
    first-time / AI-directed contributors
- docs/AGENT-CLAIM-PROTOCOL.md: fix markdownlint MD051 on
  TL;DR anchor (rename "TL;DR - five steps" ->
  "TL;DR: five steps" so GitHub slug algorithm produces a
  stable anchor; unblocks PR #108 CI)
- docs/BACKLOG.md: P2 row "First-principle seed to Zeta
  derivation - common vernacular only" per Aaron
  2026-04-22 directive

Stacked on agent-claim-protocol-git-native (PR #108) because
FIRST-PR.md references AGENT-CLAIM-PROTOCOL.md.

* docs/BACKLOG.md: P2 row — capability-limited-AI bootstrap guide

Aaron 2026-04-22: "new persona capibilty cripiled AI like small
context or few paraments how can they use us to bootstrap and
become whole guide backlog".

Filed above the first-principle-seed P2 row. Proposes
docs/CAPABILITY-LIMITED-BOOTSTRAP.md covering five capability-
limited persona shapes (short-context / small-parameter /
tool-restricted / no-web-sandbox / stateless) with a rung-by-rung
bootstrap ladder and a falsification anchor (can a 7B-parameter
model open a merged PR using only docs/FIRST-PR.md + one issue
URL?).

Composes with the vibe-coded hypothesis, AGENT-CLAIM-PROTOCOL.md
report-back mode, and the witnessable-self-directed-evolution
discipline — if crippled agents can use the factory, its
intelligence is genuinely externalised rather than large-model-
carrier-transported.

* docs/BACKLOG.md: revision — personal-resilience axis on capability-limited-AI bootstrap row

Aaron 2026-04-22 same-day extension: "in case i evver loose all my
money that bootstraping from tiny AI to become the most powerful
would sure come in handy to save money".

Adds personal-resilience as a second rationale axis alongside
capability-democratisation. The factory is meant to survive its
maintainer losing access to frontier models — if circumstances
force a move to local / open-weights / tiny models, the factory
keeps working. Pragmatic test: a ~4-8B-parameter local-inference
model on commodity hardware produces meaningful factory work when
paired with the bootstrap guide + soul-file + tests as error-
corrector.

Dated revision block preserves chronology per capture-everything
discipline. Composes with soul-file germination (already designed
to bootstrap from nothing; now extended to "from a tiny model on
a laptop").

* docs/BACKLOG.md: strip stray conflict marker from rebase resolution

…ents (one-URL handoff) (#108)

* docs: land AGENT-CLAIM-PROTOCOL.md — git-native claim spec for external agents

Standalone, linkable doc so a maintainer can hand one URL (plus the
task URL) to an external agent — ChatGPT, Codex, Gemini, Deep Research,
human contributor on another harness — and have them onboard from
zero context.

- New doc at docs/AGENT-CLAIM-PROTOCOL.md with:
  - Paste-ready hand-off template (one URL + one task + one mode-name)
  - Five-step TL;DR (clone/pull -> check -> claim -> work -> release)
  - Five mode sections: Deep Research (read-only, no claim),
    Codex sandbox (PR-write), direct-commit (maintainer),
    review/advisory (no claim), report-back / write-via-maintainer
    (no claim; maintainer commits the artifact)
  - Claim file shape (docs/claims/<slug>.md) with template
  - Slug rules, session identity, lifecycle (claim / progress / release)
  - Conflict resolution (merge = coordination), stale-claim force-release
  - Scope discipline section (claim code not identity;
    facts not framings; scope the maintainer asked for)
  - Platform adapters (GH Issues / Jira / Linear) — optional mirrors
  - "What this protocol does NOT do" block

- docs/AGENT-ISSUE-WORKFLOW.md §"The claim / lock protocol"
  now cross-references AGENT-CLAIM-PROTOCOL.md as the git-native
  substrate specification; adapter table git-native row points
  to docs/claims/<slug>.md mechanism. References section adds
  the pointer.

- AGENTS.md §"Contributor required reading" adds both
  AGENT-CLAIM-PROTOCOL.md (one-link external-agent onboarding)
  and AGENT-ISSUE-WORKFLOW.md (dual-track principle + adapter
  choice) as explicit required reading — they were only
  discoverable transitively before.

UX geared to observed external-agent responses: ChatGPT Deep Research
and Gemini review patterns both surface honest substrate-limits
("I can't push to git directly") followed by structured output
artefacts. The report-back mode is first-class so agents in that
shape have a welcome, not a gate.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

* docs/AGENT-CLAIM-PROTOCOL.md: fix markdownlint MD051 on TL;DR anchor

CI (markdownlint-cli2) flagged "[TL;DR](#tl-dr-five-steps)" as an
invalid link fragment. The GitHub slug algorithm transforms
"## TL;DR — five steps" to "tl-dr--five-steps" (double dash from
the em-dash), not "tl-dr-five-steps".

Simpler fix than chasing the slug algorithm: rename the header to
"## TL;DR: five steps" (colon strips cleanly; no em-dash) so the
slug is the predictable "tldr-five-steps". Update the link to
match. Semantics unchanged.

Unblocks PR #108 auto-merge.

* docs: FIRST-PR.md — first-class entry surface for fresh devs + vibe coders (#109)

* docs: FIRST-PR.md first-class entry surface for fresh devs + vibe coders

Aaron 2026-04-22: "can we make a first class surface for that entry
point for fresh devs and vibe coders who are just starting out".

- New docs/FIRST-PR.md promoted to first-class surface:
  - UI-first four-click contribution path (no CLI, no clone, no
    F# needed)
  - Web-UI file-edit walkthrough (GitHub pencil icon -> fork ->
    edit -> PR)
  - Vibe-coder flow (human directs AI, AI does writing, human
    reviews proposed diff)
  - Human claim walkthrough (GitHub Issues only; no git-native
    claim files required)
  - "What you do NOT need to worry about" section
- Pointers added:
  - README.md: new "Contributing" section with three entry
    points (FIRST-PR / CONTRIBUTING / AGENT-CLAIM-PROTOCOL)
  - AGENTS.md §"Contributor required reading": FIRST-PR.md
    entry above AGENT-CLAIM-PROTOCOL.md
  - CONTRIBUTING.md: top-banner pointer to FIRST-PR.md for
    first-time / AI-directed contributors
- docs/AGENT-CLAIM-PROTOCOL.md: fix markdownlint MD051 on
  TL;DR anchor (rename "TL;DR - five steps" ->
  "TL;DR: five steps" so GitHub slug algorithm produces a
  stable anchor; unblocks PR #108 CI)
- docs/BACKLOG.md: P2 row "First-principle seed to Zeta
  derivation - common vernacular only" per Aaron
  2026-04-22 directive

Stacked on agent-claim-protocol-git-native (PR #108) because
FIRST-PR.md references AGENT-CLAIM-PROTOCOL.md.

* docs/BACKLOG.md: P2 row — capability-limited-AI bootstrap guide

Aaron 2026-04-22: "new persona capibilty cripiled AI like small
context or few paraments how can they use us to bootstrap and
become whole guide backlog".

Filed above the first-principle-seed P2 row. Proposes
docs/CAPABILITY-LIMITED-BOOTSTRAP.md covering five capability-
limited persona shapes (short-context / small-parameter /
tool-restricted / no-web-sandbox / stateless) with a rung-by-rung
bootstrap ladder and a falsification anchor (can a 7B-parameter
model open a merged PR using only docs/FIRST-PR.md + one issue
URL?).

Composes with the vibe-coded hypothesis, AGENT-CLAIM-PROTOCOL.md
report-back mode, and the witnessable-self-directed-evolution
discipline — if crippled agents can use the factory, its
intelligence is genuinely externalised rather than large-model-
carrier-transported.

* docs/BACKLOG.md: revision — personal-resilience axis on capability-limited-AI bootstrap row

Aaron 2026-04-22 same-day extension: "in case i evver loose all my
money that bootstraping from tiny AI to become the most powerful
would sure come in handy to save money".

Adds personal-resilience as a second rationale axis alongside
capability-democratisation. The factory is meant to survive its
maintainer losing access to frontier models — if circumstances
force a move to local / open-weights / tiny models, the factory
keeps working. Pragmatic test: a ~4-8B-parameter local-inference
model on commodity hardware produces meaningful factory work when
paired with the bootstrap guide + soul-file + tests as error-
corrector.

Dated revision block preserves chronology per capture-everything
discipline. Composes with soul-file germination (already designed
to bootstrap from nothing; now extended to "from a tiny model on
a laptop").

* docs/BACKLOG.md: strip stray conflict marker from rebase resolution

* docs/AGENT-CLAIM-PROTOCOL.md: drain PR #108 review threads

Resolves 13 unresolved review threads on PR #108:

- Track docs/claims/ with a README.md placeholder so the
  TL;DR `ls docs/claims/` lookup works on a fresh clone.
- Hand-off template + Codex/sandbox section: declare the
  report-back fallback explicitly and require pushing the
  claim commit early (sandbox-local claims are invisible).
- Switch claim-file Session field to opaque Session ID so
  permanent git history does not carry direct agent or
  human handles (per "no name attribution" rule).
- Replace the "Amara / Gemini" example narrative with a
  role-referenced version ("external-AI Deep Research /
  architectural review").
- Correct the factually-wrong "memory/ is not in the git
  tree" claim. memory/ is tracked; the rule is "external
  claim-holders do not modify it without explicit scope."
- Reconcile docs/AGENT-ISSUE-WORKFLOW.md: the (3) git-native
  section and the claim-mechanism table now both point at
  AGENT-CLAIM-PROTOCOL.md as the authoritative substrate;
  backlog-row markers are row-local annotations, not the
  lock. Em-dash → hyphen in the table commit-message
  examples to match the canonical pattern.

Anchor link `#tldr-five-steps` was already correct
post-e5b9d14; that thread is stale.

PR-preservation drain log: docs/pr-preservation/108-drain-log.md.

* docs/pr-preservation/108-drain-log.md: append drain log

Per Otto-250 PR-preservation: per-thread (id, file:line,
verbatim original comment, outcome, verbatim reply,
resolution commit) for all 13 threads on PR #108.

Resolution SHA: 541e7b7.

* docs/AGENT-CLAIM-PROTOCOL.md: fix markdownlint MD032 (replace + with prose)

* docs: drain late-review threads on PR #108 (round 2 — 9 threads)

Address late Copilot re-review on AGENT-CLAIM-PROTOCOL,
AGENT-ISSUE-WORKFLOW, FIRST-PR, and the claims/ README.

- TL;DR step 2, AGENT-ISSUE-WORKFLOW git-native section + table
  row, and the claims/README all now describe discovery via
  `git fetch origin && git branch -r --list 'origin/claim/*'`
  plus `git show origin/claim/<slug>:docs/claims/<slug>.md`.
  `ls docs/claims/` is reframed as the merged-but-not-released
  view, not the active-claim view.
- Claim-commit-details paragraph now requires the claim commit
  to land on `claim/<slug>` specifically — that branch name
  gives push-time slug uniqueness via non-fast-forward
  rejection. Arbitrary branch names break the lock.
- Co-claim example switched from `Session: agent-A, agent-B`
  to `Sessions: sess_7f3a9c2d, sess_b84e1a6f` to match the
  no-direct-handle rule from the round-1 template rename.
- FIRST-PR.md: corrected the misleading "not in the repo"
  note about the dismissive-closing memory file (it is
  in-repo and indexed). Replaced direct reviewer names in
  three places with role refs (harsh-critic, maintainability
  reviewer, documentation agent), with the EXPERT-REGISTRY
  link as the canonical name source.

Drain log appended (no edits to the round-1 section per
Otto-229 append-only).

---------

Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com>