Composed Tier-2/Tier-3 harness and installed acceptance successor of excluded #141 #227

Open
opened 2026-10-27 21:23:03 +00:00 by kilo · 1 comment
Member

Goal

Product owner: Scarcity Router (receiving composed acceptance). Registered by the 2026-10-27 owner backlog disposition (D-083) as the active successor of excluded wontfix issue #141. Own the outstanding Tier-2/Tier-3 harness and installed integration acceptance as a downstream stage that follows the required component implementations and artifacts — never an upstream implementation prerequisite for them.

Why and Evidence

#141 (child I of program #132) was the composed acceptance/closeout child: walk the complete self-hosted path as one product, rehearse the representative target against composed fixtures, update docs coherently and deliver the program closeout report. Its D-056 pt 10 addendum defines the layered evidence shape (Tier 1 generic protocol fixtures — the #133 family, delivered; Tier 2 representative coding harness; Tier 3 broader harness evidence as practical). The 2026-10-27 reconciliation found #141 wontfix-excluded while docs/roadmap.md and docs/m10-acceptance.md still assign it composed acceptance, and found the former #186 → #141 dependency inverted relative to the canonical ordering (composed acceptance receives installed proof after component delivery). Children A–F of the program are delivered and closed; G/H/I obligations are re-homed (#139 → #224, #140 → #226 plus #183 producer state, this issue for I).

Scope

  1. Walk the full self-hosted path against the real repository: install server → configure TLS → start service/container → install worker → pair → authenticate execution source → install worker service → discover models → expose logical models → issue client key → configure OpenAI-compatible client → execute. Identify steps still requiring repository knowledge or hand-written commands; remove unnecessary friction (minimal professional open-source UX, no wizard for its own sake). Substantive product changes discovered here get their own issues.
  2. Composed rehearsal against composed fixtures at the D-056 pt 10 tiers: Tier-1 deterministic protocol fixtures; Tier-2 representative coding-harness-shaped rehearsal (bare logical model + exact effort + tools + high output budget through the public boundary; client tool round trip to the evidenced level; exact-pin still green; restricted models never selected; audit shows selected == executed); Tier-3 broader harness evidence as practical; unevidenced harnesses recorded NOT LIVE VERIFIED.
  3. Concise documentation updates across architecture, execution-surface, worker-protocol, security, release/install and acceptance docs; verify no client-identity branching in the routing core.
  4. Closeout report: issues, PRs, exact heads reviewed, gate results, external contract changes, remaining unsupported behavior, security-review state, remaining owner-side live acceptance steps, tiered evidence stated explicitly.

Acceptance Criteria

  1. End-to-end composed fixtures cover the walked path deterministically; offline only; no live inference or subscription use.
  2. Tier statements are honest: no compatibility claim without evidence; NOT LIVE VERIFIED where a harness cannot be exercised.
  3. Docs-links/integrity checks pass; the closeout report is delivered.
  4. Full gate (uv sync --only-dev, make check, uv run basedpyright, git diff --check); fresh independent technical+UX review at exact frozen HEAD/base.
  5. Live/installed owner-side acceptance remains a separately authorized, visibly open gate (LIVE_WINDOWS_ACCEPTANCE, LIVE_CODEX_SUBSCRIPTION, host diversity per docs/m10-acceptance.md); this issue does not claim it.

Constraints

Docs + friction fixes + composed tests only; substantive product changes get their own issues. No release, tag, deployment, main promotion, live provider calls or signing. This issue depends on the component implementations and artifacts it receives (native dependencies on #183 and #186); it must never block their construction.

Excluded predecessor #141 (historical, unchanged); D-083 disposition; program history #132; D-056 pt 10 layered acceptance; component producers #183 (state/events), #186 (artifacts/platform), #224 (transport/TLS), #226 (admin UX); docs/m10-acceptance.md.

## Goal **Product owner: Scarcity Router (receiving composed acceptance)**. Registered by the 2026-10-27 owner backlog disposition (D-083) as the active successor of excluded `wontfix` issue #141. Own the outstanding Tier-2/Tier-3 harness and installed integration acceptance as a **downstream stage that follows the required component implementations and artifacts** — never an upstream implementation prerequisite for them. ## Why and Evidence #141 (child I of program #132) was the composed acceptance/closeout child: walk the complete self-hosted path as one product, rehearse the representative target against composed fixtures, update docs coherently and deliver the program closeout report. Its D-056 pt 10 addendum defines the layered evidence shape (Tier 1 generic protocol fixtures — the #133 family, delivered; Tier 2 representative coding harness; Tier 3 broader harness evidence as practical). The 2026-10-27 reconciliation found #141 `wontfix`-excluded while `docs/roadmap.md` and `docs/m10-acceptance.md` still assign it composed acceptance, and found the former #186 → #141 dependency inverted relative to the canonical ordering (composed acceptance receives installed proof after component delivery). Children A–F of the program are delivered and closed; G/H/I obligations are re-homed (#139 → #224, #140 → #226 plus #183 producer state, this issue for I). ## Scope 1. Walk the full self-hosted path against the real repository: install server → configure TLS → start service/container → install worker → pair → authenticate execution source → install worker service → discover models → expose logical models → issue client key → configure OpenAI-compatible client → execute. Identify steps still requiring repository knowledge or hand-written commands; remove unnecessary friction (minimal professional open-source UX, no wizard for its own sake). Substantive product changes discovered here get their own issues. 2. Composed rehearsal against composed fixtures at the D-056 pt 10 tiers: Tier-1 deterministic protocol fixtures; Tier-2 representative coding-harness-shaped rehearsal (bare logical model + exact effort + tools + high output budget through the public boundary; client tool round trip to the evidenced level; exact-pin still green; restricted models never selected; audit shows selected == executed); Tier-3 broader harness evidence as practical; unevidenced harnesses recorded NOT LIVE VERIFIED. 3. Concise documentation updates across architecture, execution-surface, worker-protocol, security, release/install and acceptance docs; verify no client-identity branching in the routing core. 4. Closeout report: issues, PRs, exact heads reviewed, gate results, external contract changes, remaining unsupported behavior, security-review state, remaining owner-side live acceptance steps, tiered evidence stated explicitly. ## Acceptance Criteria 1. End-to-end composed fixtures cover the walked path deterministically; offline only; no live inference or subscription use. 2. Tier statements are honest: no compatibility claim without evidence; `NOT LIVE VERIFIED` where a harness cannot be exercised. 3. Docs-links/integrity checks pass; the closeout report is delivered. 4. Full gate (`uv sync --only-dev`, `make check`, `uv run basedpyright`, `git diff --check`); fresh independent technical+UX review at exact frozen HEAD/base. 5. Live/installed owner-side acceptance remains a separately authorized, visibly open gate (LIVE_WINDOWS_ACCEPTANCE, LIVE_CODEX_SUBSCRIPTION, host diversity per `docs/m10-acceptance.md`); this issue does not claim it. ## Constraints Docs + friction fixes + composed tests only; substantive product changes get their own issues. No release, tag, deployment, main promotion, live provider calls or signing. This issue depends on the component implementations and artifacts it receives (native dependencies on #183 and #186); it must never block their construction. ## Links Excluded predecessor #141 (historical, unchanged); D-083 disposition; program history #132; D-056 pt 10 layered acceptance; component producers #183 (state/events), #186 (artifacts/platform), #224 (transport/TLS), #226 (admin UX); `docs/m10-acceptance.md`.
Author
Member

Current owner triage, 2026-11-01: P2, downstream acceptance. Native dependencies #183 and #186 remain valid for receiving their producer state and installed artifacts; this issue must never block those components' implementation. Router owns composed Tier-2/Tier-3 and installed acceptance; actual component receipts and separately authorized live/platform evidence remain unmet. Earliest independent work is fixture/acceptance planning and evidence inventory without claiming missing components or live compatibility as passed. No new dependency or live execution is authorized.

**Current owner triage, 2026-11-01: P2, downstream acceptance.** Native dependencies #183 and #186 remain valid for receiving their producer state and installed artifacts; this issue must never block those components' implementation. Router owns composed Tier-2/Tier-3 and installed acceptance; actual component receipts and separately authorized live/platform evidence remain unmet. Earliest independent work is fixture/acceptance planning and evidence inventory without claiming missing components or live compatibility as passed. No new dependency or live execution is authorized.
Sign in to join this conversation.
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Reference
Creatidy/scarcity-router#227
No description provided.