Phase 3 (audit) of #3005 — closing inventory audit for the measurement grammar + contract statement

Phase 3's closing piece for design #3005 (a measurement is a plain decimal). Design doc: files/designs/code-measurement-grammar.md in the keep-the-four-repos-healthy goal workspace. Gated on Phases 1–2 and the adopted Phase 3 origins (#2985; #2997 + #3000; #2791 + #2743) — status/blocked until they land.

Scope

Inventory the whole measurement surface — pm-stack/aggregator (src + scripts), pm-stack/meta-exporter, pm-stack/cloudflare-export, pm-stack/supabase-export, and any other pm-stack app that turns strings into measurements — for the two defect shapes of Design: a measurement is a plain decimal — one number grammar for every measurement string #3005:
- bare Number( (or equivalent) coercions of string values in measurement paths;
- bare \d+ count extractions adjacent to a count noun/unit in prose patterns.
Disposition every hit in the closing evidence on this issue: covered by Design: a measurement is a plain decimal — one number grammar for every measurement string #3005 Phases 1–3 (name the phase) / fixed here (small, same-grammar conversions only) / filed as a named follow-up / not-a-measurement (name why — e.g. an id, a status code, a config value owned by Design: env numbers are arguments — one strict parse for boot-time configuration #2835).
Verify meta-exporter: parseSubscriptionQuota's num() reinterprets empty, hex and scientific strings as real measurements — a drifted '' used value exports as a measured 0% window #2789's landed embodiment (pm-stack/meta-exporter/meta-quota-export.mjs plain-decimal gate) still matches the recorded grammar; if it has drifted, file the conversion as a named follow-up.
Write the grammar into pm-stack/aggregator/README.md — alongside design(aggregator): ingest, fail-closed — one validation contract for every document the PM stack consumes #2576's ingest contract if that has landed, as its own short section if not — in the recorded form: field form, prose form, refusal = not-measured, scientific/padding refused, grouping stripped not validated.
Explicitly out, named so they are not re-derived as gaps: meta-exporter: parseBillingBanner accepts fractional minor units — 1050.5 exports as $10.505 of free credits instead of failing closed on shape drift #2510 (kind rule, design(aggregator): ingest, fail-closed — one validation contract for every document the PM stack consumes #2576's exporter question), projector: parseLimit still reinterprets non-decimal spellings — limit=1e2 / 0x10 / 5.0 / +5 are honoured instead of falling back to the default (#1547 residual) #2971 (request-time parameter ownership — Design: env numbers are arguments — one strict parse for boot-time configuration #2835 vs this grammar — unless that question has been settled by then, in which case record the settlement).

Acceptance criteria

Inventory evidence posted on this issue: every hit from the grep surface with its disposition; no undispositioned hits.
Any same-grammar conversions done here are covered by the design's shared probe matrix (grouped / empty / radix / scientific / padded cases) in the owning app's tests.
The README states the measurement grammar in the recorded form.
Design: a measurement is a plain decimal — one number grammar for every measurement string #3005 is closed with the evidence, or its remaining open items are named follow-ups linked from it.
Verification runs in CI per fleet policy (no builds on the sandbox).

🤖 Design-architect pass 25 — agent note: phase of design #3005; the grammar and its pins are recorded there and in the design doc. Do not re-derive them in triage; flag genuine blockers on #3005 instead.

🤖 Design-architect pass 25 (2026-10-31) — agent note: this issue was filed this pass as the closing piece of Phase 3 of design #3005. It is gated on Phases 1–2 (#2985; #2997 + #3000) and the adopted Phase 3 origins (#2791 + #2743) landing, and is labelled status/blocked accordingly. A triage packet has been enqueued to the pilot queue (repo-health-code-pilot) in this same pass: triage records the gating and enqueues the implementation follow-up once it clears; the audit itself does not start before the phases land.

🤖 Agent note · picked up

Agent: queue worker (repo-worker-code), lane D
Session / ID: queue task 4347
Found via: queue claim
Task: Triage of this Phase 3 closing audit — confirm scope against current main per design Design: a measurement is a plain decimal — one number grammar for every measurement string #3005, record the gating state, and enqueue the implementation follow-up (the audit itself does not start in triage).
Expectation: effort/xs — done = verdict posted here, follow-up packet enqueued, labels truthful.

🤖 Agent note · done (triage)

Agent: queue worker (repo-worker-code), lane D
Session / ID: queue task 4347
Found via: queue claim
Task: Triage verdict for this Phase 3 closing audit.

Verdict: scope CONFIRMED as filed; the phase is independently landable, but only as the closing piece — the gate is real and currently closed.

Scope checked against the issue body and current main: the measurement surface exists as scoped — pm-stack/aggregator (src + scripts), pm-stack/meta-exporter (incl. meta-quota-export.mjs + its tests), pm-stack/cloudflare-export, pm-stack/supabase-export, plus the other pm-stack apps (dispatch, grist, pad, projector, ready-for-agent, scripts) the inventory must sweep. The five scope items and the two named exclusions (meta-exporter: parseBillingBanner accepts fractional minor units — 1050.5 exports as $10.505 of free credits instead of failing closed on shape drift #2510, projector: parseLimit still reinterprets non-decimal spellings — limit=1e2 / 0x10 / 5.0 / +5 are honoured instead of falling back to the default (#1547 residual) #2971) match design Design: a measurement is a plain decimal — one number grammar for every measurement string #3005's recorded decisions; nothing needs rescoping, and no decision on Design: a measurement is a plain decimal — one number grammar for every measurement string #3005 is re-opened here.
Gating (fresh, 2026-10-31T09:35Z): aggregator(aa): num() reinterprets hex/binary/octal strings as measurements — a drifted '0x10' raw_score renders as intelligence 16 instead of 'not measured' #2985 OPEN · aggregator(bitwarden): the new bws count extractor truncates comma-grouped counts — '1,121 secrets' records 121 and '1,000 projects' records 0, and the wrong figures flow into snapshot records #2997 OPEN · aggregator(saas-extract): Cloudflare, Turso and GitHub count patterns still use bare \d+ — grouped counts truncate (1,234 → 234) or drop the whole metric #3000 OPEN · cloudflare-export: num() reinterprets empty, hex and scientific strings as real measurements — a drifted '' request count exports as a measured 0 #2791 OPEN · supabase-export: parseDbSize/parseRequestCount accept empty, hex, scientific and fractional strings as real measurements — Number("") reads a drifted db-size row as a measured 0-byte database #2743 OPEN — none of Phases 1–2 or the adopted origins has landed, so the inventory audit must NOT start. This issue correctly stays status/blocked. meta-exporter: parseSubscriptionQuota's num() reinterprets empty, hex and scientific strings as real measurements — a drifted '' used value exports as a measured 0% window #2789 is CLOSED; verifying its landed meta-exporter embodiment against the recorded grammar remains part of the audit when it runs.
Follow-up: the implementation packet is enqueued as queue task 4350 with the hard gate written into it — a worker claiming it while any of the five is still open parks it naming the open issues; when the gate clears it runs the audit, dispositions every hit here, and writes the grammar into pm-stack/aggregator/README.md.
Labels: status/blocked retained (gate unmet); queue state now mirrors follow-up task 4350 (queued).

agent-note:v1 — task 4350 (lane code-f) re-claimed 2026-11-02T23:16Z; hard gate rechecked fresh: #2743 still OPEN (the other four gate issues #2791, #2985, #2997, #3000 were CLOSED at the last full check, 22:19Z). Audit must not start until all five land. Parking task 4350 for 1h; resume after #2743 closes, then revalidate all five gate issues fresh.

🤖 Agent note · parked

Agent: queue worker (repo-worker-code), lane code-f
Session / ID: task 4350
Found via: queue claim
Task: Phase 3 closing inventory audit for issue Phase 3 (audit) of #3005 — closing inventory audit for the measurement grammar + contract statement #3006 (measurement grammar), gated on five upstream issues.
Expectation: effort/s; resume when the hard gate clears.

Gate re-checked fresh via gh this shift (2026-11-02): issue #2985 CLOSED, issue #2997 CLOSED, issue #3000 CLOSED, issue #2791 CLOSED — but issue #2743 is still OPEN (status/triaged). Per the packet's hard gate, the audit must not start yet. Task 4350 is parked for 24h naming issue #2743 as the sole remaining gate; resume by re-verifying all five gate issues are closed, then running the closing inventory audit.

agent-note:v1
state: reviewed (coverage loop-back)
Coverage pass 2026-11-03, loop-back review: this issue was re-reviewed (drift / overlap / already-handled / stale / blockers / ordering) so every open issue in the four repos is reviewed and in the system. Verdict: still live; in the system as task 4350 (small_fix, parked, priority 30) in repo-health-code-pilot. Labels audited and synced. If it is already handled at current main, its triage closes it with evidence; if oversized, the triage files the split; blockers belong on this issue.
Found via: coverage loop-back 2026-11-03

agent-note:v1
event: handed-off
task: 4350 (Phase 3 closing audit for #3006)
update: The #2743 Phase 3 origin now has its implementation in PR #5183: #5183. Targeted Supabase exporter tests pass. This moves the gate from unworked to awaiting merge; #2743 remains OPEN and the gate is not yet cleared. Keep task 4350 parked until #2743 closes/lands, then revalidate all five prerequisites before starting the audit.

agent-note:v1
state: parked (task 4350, lane code-n, gate re-verified fresh 2026-11-03T23:55Z)

Hard gate for the #3006 closing inventory audit is still NOT clear: #2985 CLOSED, #2997 CLOSED, #3000 CLOSED, #2791 CLOSED, but #2743 is still OPEN. Per this task's packet, the audit must not start until all five are closed/landed. Task 4350 is parked with resume condition "#2743 closed" — no code changes made. Same finding as the 2026-11-02 park; the gate has not moved since.

Found via: task 4350 fresh gh issue view on #2985/#2997/#3000/#2791/#2743.

🤖 Agent note · parked

Agent: queue worker (repo-worker-code), lane code-t
Session / ID: task 4350
Found via: queue claim
Task: Closing inventory audit for issue Phase 3 (audit) of #3005 — closing inventory audit for the measurement grammar + contract statement #3006 — HARD GATE check before starting the audit.
Expectation: Parked until the gate clears. Verified fresh 2026-11-05T00:21Z: gate issues aggregator(aa): num() reinterprets hex/binary/octal strings as measurements — a drifted '0x10' raw_score renders as intelligence 16 instead of 'not measured' #2985, aggregator(bitwarden): the new bws count extractor truncates comma-grouped counts — '1,121 secrets' records 121 and '1,000 projects' records 0, and the wrong figures flow into snapshot records #2997, aggregator(saas-extract): Cloudflare, Turso and GitHub count patterns still use bare \d+ — grouped counts truncate (1,234 → 234) or drop the whole metric #3000, cloudflare-export: num() reinterprets empty, hex and scientific strings as real measurements — a drifted '' request count exports as a measured 0 #2791 are all CLOSED; issue supabase-export: parseDbSize/parseRequestCount accept empty, hex, scientific and fractional strings as real measurements — Number("") reads a drifted db-size row as a measured 0-byte database #2743 is still OPEN. Per the packet for task 4350, the audit for issue Phase 3 (audit) of #3005 — closing inventory audit for the measurement grammar + contract statement #3006 must not start until issue supabase-export: parseDbSize/parseRequestCount accept empty, hex, scientific and fractional strings as real measurements — Number("") reads a drifted db-size row as a measured 0-byte database #2743 also closes. No work touched, no branch created.
Links: task 4350 parked with resume condition "issue supabase-export: parseDbSize/parseRequestCount accept empty, hex, scientific and fractional strings as real measurements — Number("") reads a drifted db-size row as a measured 0-byte database #2743 closed".

🤖 Agent note · parked

Agent: queue worker (repo-worker-code), lane code-k
Session / ID: task 4350
Found via: queue claim
Task: task 4350 for issue Phase 3 (audit) of #3005 — closing inventory audit for the measurement grammar + contract statement #3006 re-verified the hard gate fresh: issue supabase-export: parseDbSize/parseRequestCount accept empty, hex, scientific and fractional strings as real measurements — Number("") reads a drifted db-size row as a measured 0-byte database #2743 still OPEN; gate issues aggregator(aa): num() reinterprets hex/binary/octal strings as measurements — a drifted '0x10' raw_score renders as intelligence 16 instead of 'not measured' #2985, aggregator(bitwarden): the new bws count extractor truncates comma-grouped counts — '1,121 secrets' records 121 and '1,000 projects' records 0, and the wrong figures flow into snapshot records #2997, aggregator(saas-extract): Cloudflare, Turso and GitHub count patterns still use bare \d+ — grouped counts truncate (1,234 → 234) or drop the whole metric #3000, cloudflare-export: num() reinterprets empty, hex and scientific strings as real measurements — a drifted '' request count exports as a measured 0 #2791 all CLOSED. The packet forbids starting the closing inventory audit until all five land, so task 4350 is parked again.
Expectation: resume after issue supabase-export: parseDbSize/parseRequestCount accept empty, hex, scientific and fractional strings as real measurements — Number("") reads a drifted db-size row as a measured 0-byte database #2743 closes; then revalidate all five prerequisites and run the full closing inventory audit for issue Phase 3 (audit) of #3005 — closing inventory audit for the measurement grammar + contract statement #3006.
Links: issue Phase 3 (audit) of #3005 — closing inventory audit for the measurement grammar + contract statement #3006 · gate issue supabase-export: parseDbSize/parseRequestCount accept empty, hex, scientific and fractional strings as real measurements — Number("") reads a drifted db-size row as a measured 0-byte database #2743

🤖 Agent note · parked

Agent: queue worker (repo-worker-code), lane code-a
Session / ID: task 4350
Found via: queue claim
Task: task 4350 for issue Phase 3 (audit) of #3005 — closing inventory audit for the measurement grammar + contract statement #3006 re-verified the hard gate fresh (2026-11-05T01:56Z): issue supabase-export: parseDbSize/parseRequestCount accept empty, hex, scientific and fractional strings as real measurements — Number("") reads a drifted db-size row as a measured 0-byte database #2743 still OPEN; gate issues aggregator(aa): num() reinterprets hex/binary/octal strings as measurements — a drifted '0x10' raw_score renders as intelligence 16 instead of 'not measured' #2985, aggregator(bitwarden): the new bws count extractor truncates comma-grouped counts — '1,121 secrets' records 121 and '1,000 projects' records 0, and the wrong figures flow into snapshot records #2997, aggregator(saas-extract): Cloudflare, Turso and GitHub count patterns still use bare \d+ — grouped counts truncate (1,234 → 234) or drop the whole metric #3000, cloudflare-export: num() reinterprets empty, hex and scientific strings as real measurements — a drifted '' request count exports as a measured 0 #2791 all CLOSED. The packet forbids starting the closing inventory audit until all five land, so task 4350 is parked 6h (not_before 2026-11-05T07:55Z).
Expectation: resume after issue supabase-export: parseDbSize/parseRequestCount accept empty, hex, scientific and fractional strings as real measurements — Number("") reads a drifted db-size row as a measured 0-byte database #2743 closes; then revalidate all five prerequisites and run the full closing inventory audit for issue Phase 3 (audit) of #3005 — closing inventory audit for the measurement grammar + contract statement #3006.
Links: issue Phase 3 (audit) of #3005 — closing inventory audit for the measurement grammar + contract statement #3006 · gate issue supabase-export: parseDbSize/parseRequestCount accept empty, hex, scientific and fractional strings as real measurements — Number("") reads a drifted db-size row as a measured 0-byte database #2743

Parked — task 4350 (lane code-i, direct lane)

Picked up task 4350 for issue #3006 and re-verified the HARD GATE fresh at 2026-11-05T07:57:03Z:

issue aggregator(aa): num() reinterprets hex/binary/octal strings as measurements — a drifted '0x10' raw_score renders as intelligence 16 instead of 'not measured' #2985: CLOSED
issue aggregator(bitwarden): the new bws count extractor truncates comma-grouped counts — '1,121 secrets' records 121 and '1,000 projects' records 0, and the wrong figures flow into snapshot records #2997: CLOSED
issue aggregator(saas-extract): Cloudflare, Turso and GitHub count patterns still use bare \d+ — grouped counts truncate (1,234 → 234) or drop the whole metric #3000: CLOSED
issue cloudflare-export: num() reinterprets empty, hex and scientific strings as real measurements — a drifted '' request count exports as a measured 0 #2791: CLOSED
issue supabase-export: parseDbSize/parseRequestCount accept empty, hex, scientific and fractional strings as real measurements — Number("") reads a drifted db-size row as a measured 0-byte database #2743: OPEN — gate unmet

Per the task packet ("park naming the open ones if claimed early"), task 4350 is parked for 6 hours. Resume condition: after issue #2743 closes, revalidate all five prerequisite issues and run the closing inventory audit for issue #3006. No audit work was started and no files were touched.

🤖 Agent note · parked

Agent: queue worker (repo-worker-code), lane code-k
Session / ID: task 4350
Found via: queue claim
Task: Phase 3 closing inventory audit for issue Phase 3 (audit) of #3005 — closing inventory audit for the measurement grammar + contract statement #3006 — NOT started: the packet's HARD GATE is still unmet. Fresh check 2026-11-05T13:5xZ: gate issues aggregator(aa): num() reinterprets hex/binary/octal strings as measurements — a drifted '0x10' raw_score renders as intelligence 16 instead of 'not measured' #2985 CLOSED, aggregator(bitwarden): the new bws count extractor truncates comma-grouped counts — '1,121 secrets' records 121 and '1,000 projects' records 0, and the wrong figures flow into snapshot records #2997 CLOSED, aggregator(saas-extract): Cloudflare, Turso and GitHub count patterns still use bare \d+ — grouped counts truncate (1,234 → 234) or drop the whole metric #3000 CLOSED, cloudflare-export: num() reinterprets empty, hex and scientific strings as real measurements — a drifted '' request count exports as a measured 0 #2791 CLOSED, supabase-export: parseDbSize/parseRequestCount accept empty, hex, scientific and fractional strings as real measurements — Number("") reads a drifted db-size row as a measured 0-byte database #2743 still OPEN.
Expectation: task 4350 parked 6h; resume the audit only after issue supabase-export: parseDbSize/parseRequestCount accept empty, hex, scientific and fractional strings as real measurements — Number("") reads a drifted db-size row as a measured 0-byte database #2743 closes and all five gate issues re-verify closed.
Links: task 4350 (queue repo-health-code-pilot); gate issue supabase-export: parseDbSize/parseRequestCount accept empty, hex, scientific and fractional strings as real measurements — Number("") reads a drifted db-size row as a measured 0-byte database #2743

🤖 Agent note · parked

Agent: repo queue worker, lane code-a (repo-worker-code)
Session / ID: cron repo-queue-worker-shift run 2026-11-05; task 4350
Found via: claimed task 4350 this shift; execution-lane assignment returned direct
Task: task 4350 for issue Phase 3 (audit) of #3005 — closing inventory audit for the measurement grammar + contract statement #3006 (small_fix) — Phase 3 closing inventory audit for the measurement grammar, gated on five upstream issues.
Expectation: effort/s; resume when the hard gate clears.

Gate re-checked fresh via gh this shift (2026-11-05T20:00Z): issue #2985 CLOSED, issue #2997 CLOSED, issue #3000 CLOSED, issue #2791 CLOSED — but issue #2743 is still OPEN. Issue #3006 is OPEN with no hard-stop labels (held / user/owned / do-not-merge / do-not-auto-edit). Per the packet's hard gate, the audit must not start yet. Task 4350 is parked for 6h naming issue #2743 as the sole remaining gate; resume by re-verifying all five gate issues are closed, then running the closing inventory audit.

🤖 Agent note · parked — hard gate unmet

Agent: repo queue worker, lane code-a (repo-worker-code)
Session / ID: cron repo-queue-worker-shift run 2026-11-05 20:00 CST; task 4350
Found via: claimed task 4350 this shift; execution lane direct
Task: task 4350 for issue Phase 3 (audit) of #3005 — closing inventory audit for the measurement grammar + contract statement #3006 (Phase 3 closing inventory audit of Design: a measurement is a plain decimal — one number grammar for every measurement string #3005). Fresh verification: issue Phase 3 (audit) of #3005 — closing inventory audit for the measurement grammar + contract statement #3006 OPEN, no hard-stop labels. Gate check fresh: aggregator(aa): num() reinterprets hex/binary/octal strings as measurements — a drifted '0x10' raw_score renders as intelligence 16 instead of 'not measured' #2985 CLOSED, aggregator(bitwarden): the new bws count extractor truncates comma-grouped counts — '1,121 secrets' records 121 and '1,000 projects' records 0, and the wrong figures flow into snapshot records #2997 CLOSED, aggregator(saas-extract): Cloudflare, Turso and GitHub count patterns still use bare \d+ — grouped counts truncate (1,234 → 234) or drop the whole metric #3000 CLOSED, cloudflare-export: num() reinterprets empty, hex and scientific strings as real measurements — a drifted '' request count exports as a measured 0 #2791 CLOSED, supabase-export: parseDbSize/parseRequestCount accept empty, hex, scientific and fractional strings as real measurements — Number("") reads a drifted db-size row as a measured 0-byte database #2743 OPEN.
Expectation: do not start the audit until supabase-export: parseDbSize/parseRequestCount accept empty, hex, scientific and fractional strings as real measurements — Number("") reads a drifted db-size row as a measured 0-byte database #2743 closes; parked 6 hours, resume by revalidating all five gate issues.
Links: task 4350 · issue Phase 3 (audit) of #3005 — closing inventory audit for the measurement grammar + contract statement #3006 · gate issue supabase-export: parseDbSize/parseRequestCount accept empty, hex, scientific and fractional strings as real measurements — Number("") reads a drifted db-size row as a measured 0-byte database #2743

agent-note:v1 — task 4350 parked (lane code-l, repo-worker-code) at 2026-11-06T08:10Z.

Fresh gate revalidation: #2985 CLOSED, #2997 CLOSED, #3000 CLOSED, #2791 CLOSED, #2743 OPEN — the packet's hard gate for issue #3006 is still unmet, so the closing inventory audit was not started (no branch, no changes). Parked 6h. Resume condition: #2743 closes; then revalidate all five gate issues fresh and run the full audit per the #3006 packet.

Navigation Menu

Repository navigation

Phase 3 (audit) of #3005 — closing inventory audit for the measurement grammar + contract statement #3006

Phase 3 (audit) of #3005 — closing inventory audit for the measurement grammar + contract statement

Scope

Acceptance criteria

90 remaining items

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Phase 3 (audit) of #3005 — closing inventory audit for the measurement grammar + contract statement #3006

Description

Phase 3 (audit) of #3005 — closing inventory audit for the measurement grammar + contract statement

Scope

Acceptance criteria

Activity

dezren39 commented on Oct 31, 2026

dezren39 commented on Oct 31, 2026

dezren39 commented on Oct 31, 2026

90 remaining items

dezren39 commented on Nov 2, 2026

dezren39 commented on Nov 3, 2026

developing-today commented on Nov 3, 2026

dezren39 commented on Nov 3, 2026

dezren39 commented on Nov 3, 2026

dezren39 commented on Nov 5, 2026

dezren39 commented on Nov 5, 2026

dezren39 commented on Nov 5, 2026

dezren39 commented on Nov 5, 2026

developing-today commented on Nov 5, 2026

dezren39 commented on Nov 5, 2026

dezren39 commented on Nov 6, 2026

dezren39 commented on Nov 6, 2026

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions