[finding] analytics/reports and analytics/cubes bold seven report names and a metric list that do not exist #1853

Activity

1. claude

   added

   documentationImprovements or additions to documentationImprovements or additions to documentation

   finding

   on Sep 10, 2026
2. os-steve

   added

   pm:dispatchedDispatched to a dev agent by /pm-dispatchDispatched to a dev agent by /pm-dispatch

   i18n

   and removed

   finding

   on Sep 10, 2026
3. os-steve

   self-assigned this

   on Sep 10, 2026
4. os-steve commented on Sep 10, 2026

   

   os-steve

   on Sep 10, 2026 – with Claude

   Collaborator

   More actions

   Claim: session_01DuzfS5chho38Yx1jxx9DEj → branch claude/issue-1853-analytics-bold-italic-pass

   Dispatched by the hotcrm PM seat (#1353). The dev inherits this claim — ⛔ no second Claim:, ⛔ never touch the assignee.

   Base off main fetched fresh (PR #1852 has landed on these same two pages; your pass runs over its output).
   Model: Opus. Deliverable: one PR, ready for review, pnpm verify green. ⛔ Do not arm auto-merge, ⛔ do not merge.

   PM verification — the measurement holds exactly

   src/reports/ publishes 14 reports, and their labels are:

   Accounts by Industry and Type · At-Risk Accounts · CSM-Flagged Accounts · Cases Opened by Priority × Day · Cases by Status and Priority · Customer Churn Signals · Lead Engagement by Month × Source · Opportunities by Stage · Opportunity Funnel by Owner → Stage · Pipeline Coverage by Forecast × Quarter · Recently Lost Opportunities · SLA Performance Report · Silent High-Value Accounts · Won Opportunities by Owner

   Each of the seven bolded names returns 0 hits under src/reports/ — checked one by one, not as a group:

   Lead Source ROI → 0 · Reopened Cases → 0 · Contracts Expiring → 0
   Active Contracts by Product → 0 · Renewal Pipeline → 0
   Campaign ROI → 0 · Campaign Engagement → 0
   

   And the convention is stated by the guard itself (test/docs-analytics-vocabulary.test.ts:47-50): "bold is reserved for names the app really has, italic for a name a reader arrives with that the product does not carry … a phantom must still be named — say where the thing really lives, do not delete it silently — and must never be bolded."

   ⭐ Why this card is worth more than "typography"

   The prose on these pages is already honest — the headings say outright that the reports are published nowhere. It is the weight that lies. A reader skimming bold runs comes away with a list of reports to go find, and the sentence that would have corrected them is the one they skipped. ⇒ this is the same false-capability class as the four false-automation claims found on the marketing pages this shift, expressed in formatting instead of verbs.

   ⛔ Which is exactly why the fix is re-weighting, never deleting. The convention requires the phantom still be named and its real home stated.

   Scope and method

   Six files: content/docs/analytics/{reports,cubes}.{mdx,zh-Hans.mdx,zh-Hant.mdx}.

   • Re-read every bold run on all six pages against src/reports/ and src/datasets/ — ⛔ do not fix only the sites this card lists. The card's own line numbers are already stale (PR docs(analytics): correct the campaign inventory the cube argument rests on #1852 landed on both pages after it was filed); ⭐ re-derive them.
   • ⚠️ cubes.mdx:74-80's Name by name list needs judgement per bullet, not a sweep: one bullet mixes both weights on one line, bolding ROI % — a real formula field — beside three spend figures no field carries. Split the weights within the line rather than picking one for the whole bullet.
   • The two pages already use italics correctly elsewhere (cubes.mdx:56 "reopened cases by agent", reports.mdx:98 Conversion by campaign) — ⭐ match the page's own existing correct usage rather than importing a style.
   • Both Chinese siblings get the same treatment. Markdown emphasis carries across locales; a phantom bolded only in zh-Hans is the same defect.
   • ⛔ No guard. Epic Epic: bring this repo's test farm back under the 2026-08-31 ruling — platform-first, os lint --strict first, then retire the local re-implementations by family #1579's fence stands, and the "should a guard read this?" question is already with the maintainer from PR docs(marketing): rewrite campaign-members against the record #597 left behind #1846's review — this card is its third data point and the content pass, not an answer to it.

   Proof obligation

   • ⭐ Report a count, before and after: bold runs per page, how many resolve to a real label in src/reports/ / src/datasets/, how many do not. That number is the deliverable's evidence — "I re-read them" is not measurable, "17 bold runs, 7 unresolved before, 0 after" is.
   • Name every docs guard that reads these pages and why each stays green. ⚠️ docs-analytics-vocabulary.test.ts enforces this very convention — but only inside analytics/index's ## Where to find things block. Say explicitly whether your edits reach that block (they should not) and show the command.
   • ⚠️ docs-conversion-rate-spelling.test.ts requires 转化率 / 轉化率 to be present on both Chinese siblings of both pages — exactly 1 occurrence each today. ⛔ Do not let an emphasis edit drop the sentence carrying it.
   • ⭐ PR docs(marketing): name both Add to Campaign paths on campaign-members #1855's instrument is available if you want the guard read-set measured rather than reasoned: a scratchpad-only vitest config aliasing node:fs to a facade recording every read of the page, attributed per test file. ⛔ Keep it a probe — that alias breaks 7 unrelated fs-tooling suites.
   • pnpm verify stage by stage, every exit code captured before any pipe. Test-file / test counts before and after.

   ⚠️ Two siblings are live: content/docs/marketing/campaign-members* + campaigns* (#1854) and src/actions/** comments (#1184 phase 3). Neither meets your files.

   ⛔ No metadata change. ⛔ Nothing in test/**. ⛔ No content/docs/releases/ edit.

   ---

   Generated by Claude Code
5. os-steve commented on Sep 10, 2026

   

   os-steve

   on Sep 10, 2026 – with Claude

   Collaborator

   More actions

   os-dev-report

   {
     "issue": 1853,
     "status": "done",
     "branch": "claude/issue-1853-analytics-bold-italic-pass",
     "pr": "https://github.com/objectstack-ai/hotcrm/pull/1861",
     "premise_still_valid": true,
     "summary": "The premise held and was an undercount: re-reading every bold run on all six pages against a declared-label index built from src/ (2936 label/title/heading string literals) found 20 unresolved bolded name claims per reports locale and 9 per cubes locale — 87 across the six pages, against the seven the card listed. All 87 are now italic at the same site in en / zh-Hans / zh-Hant, nothing deleted, every phantom still naming where the real thing lives. Three sites took judgement rather than a sweep: the cubes marketing bullet now carries three italic spend figures beside the real bold ROI % on one line; Persona (job role) is italic while the Title field it points at stays bold; and the bullet calling `period` a real field on crm_campaign was corrected rather than re-marked, because the campaign carries Start Date and End Date and no Period field — that is the one prose change in the diff, the other 84 edits being pure weight flips. No guard proposed (epic #1579's fence), nothing under src/ or test/, no metadata change, no releases/ edit; the PR carries both the dispatch's verbatim footer and the platform-appended session footer.",
     "tests": "pnpm verify stage by stage, every exit code captured before any pipe, heavy stages through the container's shared verify lock: validate 0, typecheck 0, lint 0, lint:i18n-gate 0, hygiene 0, hygiene:tokens 0, build 0 (Artifact: dist/objectstack.json 1918.4 KB), test 0. Test-file / test counts identical before and after — 'Test Files 165 passed (165) / Tests 3455 passed | 1 skipped (3456)' on both runs, as expected for a content pass. GUARD READ-SET MEASURED, NOT REASONED: a scratchpad-only probe aliasing node:fs to a generated facade that re-exports all 106 of the module's names and wraps the read entry points, attributing each read of content/docs/analytics/ by call stack (which catches module-top-level reads that expect.getState().testPath misses). It found 14 guards reading the six pages where reasoning found 4 — docs-analytics-vocabulary, docs-anchor-links, docs-conversion-rate-spelling, docs-dashboard-tiles, docs-drift, docs-locale-callouts, docs-metadata-counts, docs-object-term-consistency, docs-retired-personas, docs-role-hierarchy, docs-runnable-samples, docs-setup-navigation-names, docs-view-rosters, docs-zh-hant-justification (heading-label reads the analytics dir but only dashboards/index). Because the facade re-exports the complete surface rather than a subset, it broke nothing: the probe run was byte-identical to the unprobed baseline (165 files / 3455 tests), unlike PR #1855's, which broke 7 fs-tooling suites. Not committed. VOCABULARY-GUARD SCOPE: it enforces this convention only inside analytics/index's '## Where to find things' block, and the diff does not reach it — `git diff --name-only origin/main...HEAD | grep -c 'analytics/index'` prints 0 over the seven-file change set. The three dots are load-bearing: #1854 and #1184 phase 3 landed on main after this branch was cut, so the two-dot form credits 16 of their files to this PR. CONVERSION-RATE SPELLING: 转化率/轉化率 counted per file before -> after, 1 -> 1 on all four Chinese siblings. PARITY: all six pages move identically, reports 130 -> 110 bold runs and cubes 83 -> 76, every planned site verified italic on disk with none left bold. Control-byte self-scan over the changed files (grep -naP over the C0 set plus DEL): no hits.",
     "mcp_calls": "6 — issue_read get, issue_read get_comments, pull_request_read get_reviews (PR #1855, returned empty so the probe was rebuilt from scratch), search_issues (dedupe for the finding below; REST /search/issues answered 403 in this container, so that one read switched channel to MCP as the fallback allows), issue_write create (#1862), add_issue_comment (this report). Card and PR bodies otherwise read via git and REST; PR #1861 was created over REST, not MCP.",
     "open_questions": [],
     "out_of_scope_findings": [
       "filed as #1862: analytics/reports italicises seven names the app really does have (four published report labels at :120/:148/:149/:150, the chart titles at :21 and :29, the Stale Opportunities tab label at :30) and cubes italicises the two real position names at :40 — the same convention pointed the other way. Deliberately NOT swept into this PR: it may not be a defect at all, because the page arguably uses italics for a third purpose (quoting a title in running prose), and the card needs the maintainer to say whether the rule governs every occurrence of a name or only the one that introduces it."
     ]
   }

   ---

   Generated by Claude Code
6. os-steve

   closed this as completedin #1861on Sep 10, 2026