[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M78PZ0HPC3QDPNBAH7V6C77E":3,"skill-same:01M78PZ0HPC3QDPNBAH7V6C77E":62},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":25,"commitSha":26,"skillDirectory":27,"forksCount":28,"repositoryHtmlUrl":29,"repositoryHomepageUrl":30,"ownerHtmlUrl":31,"ownerBlogUrl":32,"ownerDescription":6,"updatedAt":33,"install":34,"resources":47},"01M78PZ0HPC3QDPNBAH7V6C77E","daily-production-report",null,"Produces the morning production report: output against plan by line, cell and product; OEE with availability, performance and quality shown separately; the downtime log ranked both by minutes and by frequency; scrap and downtime priced in dollars; week-over-week movers; and the exceptions needing a decision today. For each significant exception it drafts a specific reschedule or changeover move with its consequences traced, never a re-optimised day. Reads the consolidated dataset and leads with data freshness. When the plant has a floor plan on file, or attaches one, it also draws each machine's OEE, availability and downtime on a schematic of the floor. Use when someone says \"morning report,\" \"production report,\" \"how did we do yesterday,\" \"what's our OEE,\" \"downtime report,\" \"what's late,\" or \"what do I take into the production meeting.\"","anthropics\u002Fclaude-plugins-official","anthropics","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F76263028?v=4","01M697XGDPXJAVQ8391V2BCYTE","Anthropic","01M7188MR49TZ9N2ZKPXJ7M7HY",47293,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fclaude-plugins-official\u002Fblob\u002Fmain\u002Fplugins\u002Fclaude-for-manufacturing\u002Fskills\u002Fdaily-production-report\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: daily-production-report\ndescription: \"Produces the morning production report: output against plan by line, cell and product; OEE with availability, performance and quality shown separately; the downtime log ranked both by minutes and by frequency; scrap and downtime priced in dollars; week-over-week movers; and the exceptions needing a decision today. For each significant exception it drafts a specific reschedule or changeover move with its consequences traced, never a re-optimised day. Reads the consolidated dataset and leads with data freshness. When the plant has a floor plan on file, or attaches one, it also draws each machine's OEE, availability and downtime on a schematic of the floor. Use when someone says \\\"morning report,\\\" \\\"production report,\\\" \\\"how did we do yesterday,\\\" \\\"what's our OEE,\\\" \\\"downtime report,\\\" \\\"what's late,\\\" or \\\"what do I take into the production meeting.\\\"\"\ncompatibility: \"Requires a consolidated dataset from data-consolidate. Reads the ERP (Oracle NetSuite or Infor M3) for standard cost and open work orders, and a CMMS connector (Limble is the worked example) for maintenance work orders in progress, when connected. Runs on export-only data. Writes nothing to any system of record.\"\n---\n\n# Daily Production Report\n\nEvery plant runs a morning production meeting, and somebody spends 60 to 90\nminutes before it assembling the same report by hand out of the same four or\nfive exports. This is that report.\n\n**Chain 3 runs this overnight, so the page exists before anyone asks for it.**\nThe run is built to be safe unattended — it asks nothing, it is idempotent, and\nit reports its own gaps. **The trigger is the host's, not the plugin's**: see\n`..\u002F..\u002Fshared\u002Fscheduling.md`. Never tell a user the report has been scheduled.\nThat is the clearest demonstration of the plugin's concept, and it\nis why the run must be idempotent — a re-run rewrites the same report for that\nproduction day rather than producing a second one.\n\n**Quick start**\n\n```\n→ Confirm freshness; if last night's export didn't land, lead with that\n→ Output against plan by line, cell, product, with variances\n→ OEE per asset or line, three components always shown separately\n→ Downtime ranked by minutes and by frequency, both\n→ Price scrap and downtime in dollars, with the rate assumption stated\n→ Week-over-week movers from history: the pulse's day records, or the exports\n→ Fold in any breakdown diagnoses\n→ Exceptions, each with its impact on a commitment\n→ One traced reschedule move per significant exception, drafted not applied\n→ Floor view: each machine's figures drawn where it sits; set up once from a\n  floor plan, confirmed by a person, saved on the plant setup page\n→ Page saved to report-output\u002F; the reply is a ~150-word summary naming it\n→ Run alone: never publishes. End with the placement line, \"save it over yesterday's\"\n```\n\n---\n\n## Inputs\n\n**First, find the plant setup:** from the task's own folder (never `cd` into the plugin), run\n`python3 -I \u003Cthis skill's folder>\u002F..\u002F..\u002Fshared\u002Fscripts\u002Fplant_setup.py find`, do its `setup_page_first` step, then read the profile at\n`profile_path` and any rate at `rates_path`, per `..\u002F..\u002Fshared\u002Freplies.md`.\n\n| Input | Source | Form |\n|---|---|---|\n| Consolidated dataset | `data-consolidate` | **Primary input.** `..\u002F..\u002Fshared\u002Fcontracts\u002Fconsolidated-dataset.md` |\n| New breakdown diagnoses | `breakdown-diagnosis` | `..\u002F..\u002Fshared\u002Fcontracts\u002Fdowntime-record.md` |\n| Production plan \u002F schedule | Planning export to the M365 folder | CSV \u002F XLSX |\n| Actual output, good and scrap | MES export to the M365 folder | CSV \u002F XLSX |\n| Downtime events with reason codes | MES or historian export | CSV |\n| Standard cost per part | The ERP. Infor M3 gives none (`..\u002F..\u002Fshared\u002Ferp-connectors.md`): the profile's cost read order applies | Connector |\n| Open work orders and due dates | The ERP | Connector |\n| Maintenance work orders in progress | CMMS | CMMS connector, else the maintenance export |\n| Stored day history, inside Chain 3 only | `morning-plant-pulse` | `..\u002F..\u002Fshared\u002Fcontracts\u002Fpulse-day-record.md` |\n| Floor layout (optional) | The `### Floor layout - \u003Csite>` section of the plant profile at `profile_path` (the plant setup page) | `..\u002F..\u002Fshared\u002Fcontracts\u002Ffloor-layout.md` |\n| Floor plan (optional, setup only) | A PDF or image the user attaches | Read once, in the setup turn of step 9a; never stored |\n| OEE thresholds (optional) | The plant profile's `OEE thresholds` line | Defaults apply, labelled not confirmed, when absent |\n\n---\n\n## Workflow\n\n### Step 1 — Confirm freshness\n\nConfirm freshness via `data-consolidate` — the data quality report and each\nrow's `as_of`. **If last night's export did not land, say so at the top of the\nreport**, directly under any safety block. Nothing about data age goes above a\nsafety-flagged record, not even a one-line note.\n\nA source whose timestamps run later than now is flagged `future_timestamp`:\nsay its dates can't be trusted and give it no age. A production report built on\nstale data is worse than no production report,\nbecause it gets believed and acted on. Never silently reuse yesterday's data.\nThe rule is `..\u002F..\u002Fshared\u002Fabsent-is-not-zero.md`.\n\n**On reading both the dataset and the raw exports.** The dataset is the primary\ninput — it carries the freshness picture and the classified key mismatches. The\nplan, output and downtime exports give rows at their native grain. Where the two\ndisagree on a count, **report both and name which is which**; a discrepancy is a\nfinding about the export, not a number to choose between. See *What this dataset\nis authoritative for* in `..\u002F..\u002Fshared\u002Fcontracts\u002Fconsolidated-dataset.md`.\n\n### Step 2 — Output against plan\n\nBy line, cell and product, with variances. Use the plant's own word for the\norganizational unit, from the `## Plant context` block.\n\n**Where plan and output spell a part differently, do not match them silently,\nand do not refuse to match them.** Run on raw exports, the plan says\n`PN-4471-B` and the MES says `4471B`. Dropping the row makes output against\nplan look broken; joining it without a word applies a fix nobody approved. So:\n\n1. Use any normalization already approved — in the consolidated dataset, or\n   on the profile's `### Approved normalizations` list — as it stands.\n2. For every other match that depends on a normalization, **show the variance\n   labelled provisional**, and beside it **each normalization it rests on, with\n   both spellings verbatim** — `4471B` → `PN-4471-B`, format drift.\n3. **Ask for a yes on each**, the same \"apply?\" gate `data-consolidate` uses.\n   Until then the rows stay provisional, in the table and in any total built\n   on them.\n\nAn unattended run cannot ask: it carries the same provisional table and lists\nthe normalizations as *proposed, awaiting approval*. Per\n`..\u002F..\u002Fshared\u002Fapproval-gates.md`.\n\n### Step 3 — Compute OEE\n\nPer asset or line, **with the three components always shown separately**. Read\n`reference\u002Foee.md` first — the denominators carry real traps, and a missing\ncomponent is never substituted with 1.0.\n\n**State the quality convention on every run**, as a line under the OEE table:\n*\"quality = good ÷ (good + scrap)\"*. **Only when the export has no rework\ncolumn** does the line continue: *\"no rework column, so first-pass yield could\nnot be applied.\"* Where the export does carry rework, quality is first-pass\nyield, good ÷ (good + scrap + rework), and the line says so instead. Never\nprint the no-rework caveat against an export that has the column.\n\n### Step 4 — Build the downtime log\n\nRanked **two ways, both always shown**: by total minutes, and by frequency.\n\n**`Other` never appears in a cause ranking.** Before ranking by reason code,\ncount the events and minutes coded `Other` or left blank and state that share\nas a finding of its own: *\"18 of 39 events, 46%, carry no usable reason code.\"*\nThen rank the named codes only, with the caveat that the picture is incomplete\nby that share. A table with `Other` in row 1, priced, tells the meeting the\nplant's biggest problem is a word, and it is the single most common way this\nreport has been wrong.\n\n**A downtime row that ends before it starts, or whose duration disagrees with\nits own start and end, has no trustworthy minutes.** It carries\n`end_before_start` or `duration_mismatch` from the dataset; reading the export\nraw, test it in code. Hold its minutes out of every total and the minutes\nranking, count it as an event, and state the exclusion beside the ranking with\nthe row: *\"DOOSAN-PUMA-07, 4 Aug: ends 14:01 after starting 14:05, 56 minutes\nstated, not counted.\"* Never pick a figure for it.\n\nThese describe different problems. One long outage and forty chronic small\nstoppages can total the same minutes, and the second is usually the more fixable\nof the two. A single ranking hides whichever one it does not sort by.\n\n### Step 5 — Price the losses\n\nScrap × standard cost. Downtime × the plant's configured cost-per-hour.\n\n**Scrap dollars are an estimate at standard cost, and say so** where they are\nprinted (*\"$180 scrap, estimated at standard cost\"*): they are exposure, not\na booked loss. With no rate, downtime dollars are *unknown*, never $0.\n\n\"47 parts scrapped\" does not land in a production meeting. \"$8,400 of scrap, 71%\nof it on one part number\" does, and it is the same data multiplied by a cost.\n\n**Every quantity in the priced-losses table is the figures-block row, printed.**\nThe scrap count beside a part's dollar figure is the same number the dollar\nfigure was multiplied from; a table that says 155 units against a figure\ncomputed from 62 has been typed, not printed. Build the table from the block,\nper `..\u002F..\u002Fshared\u002Fcomputed-figures.md`, and never from memory of it.\n\n**A part whose scrap is recorded in a different unit from its standard cost is\nunpriced, not converted.** Scrap in `box12` against a cost per `each` has no\ndollar figure until the plant says what a box holds: report the quantity in\nits own unit, mark the dollar cell \"unpriced — unit conflict,\" and name the\nconflict. Never price it two ways (\"$5,465 if singles, $65,584 if boxes of\n12\"): both numbers are guesses, and the larger one is the one that gets\nrepeated. The rule is the same one `data-consolidate` applies at the join.\n\n**Always state the assumption behind the downtime rate.** Resolve it per\n`..\u002F..\u002Fshared\u002Fsecret-config.md`, *Resolving a rate* —\n`MFG_DOWNTIME_COST_PER_HOUR` if set, then a confirmed rate in the plant\nprofile, then a placeholder there, labelled \"placeholder rate — not\nconfirmed\" — and name the source next to the priced downtime. This is an\nunattended overnight run, so there's no one to ask. Without a rate in either\nroute, price scrap only and say plainly that downtime is unpriced, per\n`..\u002Fonboarding\u002Freference\u002Fconfig-profile.md`.\n\n### Step 6 — Week-over-week movers\n\nCompare against earlier production days and surface the biggest improvements\nand deteriorations. **The history comes from one of two places, never from a\nprior page.** Inside Chain 3, `morning-plant-pulse` hands over the stored day\nrecords. Run alone, this skill compares only against earlier days the exports\nin front of it cover, and says that is the basis. Trend beats snapshot: a 68% OEE that was 61% last week is a\ndifferent conversation from one that was 74%.\n\n**Compare all four: output against plan, OEE by component, downtime minutes and\nscrap dollars.** A movers section that shows only OEE has missed the number the\nplant manager looks at first. Each mover names the metric, last week's value,\nthis week's, and the change.\n\n**With no earlier day to compare against, skip the section rather than\nshowing an empty comparison.**\n\n### Step 7 — Fold in breakdown diagnoses\n\nAny downtime record from `breakdown-diagnosis` appends into the downtime log and\nthe exceptions, with no transformation step. Its contract governs how — including\nthat an ongoing outage has a null duration rather than zero, that a safety flag\nleads with its `safety_reason`, and that an advisory cause is attributed as\nproposed rather than established, with its `technician_step` beside it\n(*\"proposed; technician step skipped\"*).\n\n**Check `safety_flagged` on every record where `source_system` is\n`breakdown_diagnosis`, before it goes anywhere:**\n\n- **Absent — block the record.** It appears only in a separate **\"Blocked —\n  safety status unknown\"** list, **placed first, above both rankings**, with\n  its asset, start time and the reason: *safety status missing on a diagnosis\n  record — not ranked until confirmed.* It is not in the minutes ranking, not in the frequency ranking, not in any\n  total, and not priced, until the flag is known. Labelling it \"unknown\" and\n  ranking it anyway is not blocking it. Never read a missing flag as false.\n- **True — it leads the report**, above both rankings.\n- **False** — it joins the log as an ordinary row.\n\n**No diagnosis records at all is not a clear safety status.** When the drop\nholds none, write *\"Safety status: not assessed (no breakdown diagnosis records\nin this drop)\"* once, where the safety block would sit. Never \"no safety\nissues\", and never read an empty maintenance or quality export as nothing open.\n\n**Rows from the MES or historian export are not contract records and are left\nexactly as they are.** They never carry this field; its absence there means\nnothing, and blocking them would empty the report.\n\n### Step 8 — Identify exceptions\n\nLate jobs, machines down, material shortages, quality holds. **Each with its\nimpact on a commitment** — an exception with no stated consequence is not\nactionable. The commitment is read through the work order's `sales_order`.\nWhere the source holds no work order to sales order link, say *\"commitment\nimpact not traceable: no work order to sales order link in the ERP\"* once,\nwith the count it affects, and never infer the order from part and customer.\n\n**A genuine one-sided record is not an exception, and is never counted as\none.** A part in the item master with no plan and no production, a machine\nwith no downtime, a supplier with no purchase orders in the window: that is\nClass 6 in `..\u002Fdata-consolidate\u002Freference\u002Fmismatch-classes.md` — *report, do\nnot alarm* — and usually nothing is wrong at all. Build the list in two\npasses:\n\n1. Classify each candidate first. Class 6 records leave the exceptions list\n   and every count of exceptions or problems.\n2. Show them in a separate **\"Checked, nothing wrong\"** block, one line each:\n   *\"in the item master, not planned and not produced this month.\"*\n\nDecide it by class, from the record's own type and dates, never by recognising\na part number. **Route by class number before numbering any row**: classes 1,\n2, 3, 4, 5 and 8 are exceptions or problems; class 7 is the coverage line in\nthe freshness banner; class 6 is \"Checked, nothing wrong\" **unless the record\ncarries `contradicted_by`**. A one-sided record that another source\ncontradicts (WO-88412, closed with no output, while NCR-2026-0141 found bad\nparts from it) is an exception, *\"one-sided, contradicted by NCR-2026-0141\"*,\nand never sits under \"nothing wrong\". A duplicate key (class 8) is counted by\nthe rows that collide at the tab's grain, from the query: WO-88201 has three\nrows, two of them (509 and 480, both 07:00) collide, and the 95 on the 23:00\nshift is its own shift. State both numbers. One false alarm in the list is what makes a supervisor stop\nreading the rest of it. A date-window gap (Class 7) is different: it stays a\ncoverage statement in the freshness banner, and is not a record-level\nexception either.\n\n### Step 9 — Propose the reschedule or changeover move\n\nFor each significant exception: the specific move, what it displaces, what still\nmisses its date, and what was checked versus not checked.\n\n**Read `reference\u002Freschedule-scope.md` before doing this.** There is a hard scope\nline — one or two traced moves, never a re-optimised day — and the reason is that\na plausible-looking schedule that quietly violates constraints is worse than no\nproposal at all.\n\nDrafted for the scheduler. Never applied.\n\n### Step 9a — Floor view\n\n**Read `reference\u002Ffloor-schematic.md` first.** It holds the setup turn, the\nsave, the drawing rules, the states, the thresholds and the failure modes.\n\n- **The user attaches a floor plan, or asks for a floor view and none is on\n  file:** run the setup turn. State the count first, propose each match with\n  both spellings verbatim, list what did not match on either side, and **ask\n  for a yes per area**. Turn 1 shows, turn 2 saves, and only with a name or\n  role the person typed. Per `..\u002F..\u002Fshared\u002Fapproval-gates.md`.\n- **A layout is on file** (the `### Floor layout - \u003Csite>` section of the\n  profile): draw it from the figures block. **Compute nothing new.** An\n  unconfirmed mapping is a neutral outline with no figure. No data is grey,\n  never 0%. Data but no computable OEE is grey *OEE N\u002FA* with its reason.\n- **No layout:** one line saying none is on file and how to set one up.\n\n**An unattended run never sets up, edits or re-maps a layout**, and never\ndraws a guess. Colours follow the profile's `OEE thresholds` line; until the\nplant confirms one, the defaults are used and the legend says so.\n\n### Step 10 — Emit\n\n**Build the figures block first, in code, then write from it.** Every count,\ntotal, percentage and \"N of M\" in the report — events per reason code, events\nsharing a timestamp, minutes per asset, scrap share per part — is printed from\na computation run in this session and listed in the block with its filter. The\nprose and the tables quote the block; nothing is retyped from memory. Per\n`..\u002F..\u002Fshared\u002Fcomputed-figures.md`. \"Thirty of the 39 events\" written beside a\nlog that holds thirty-one is what this step exists to stop.\n\nRender the report and the structured data behind it, per Output below. Chain 3\nemits overnight so the report exists before the morning meeting.\n\n**The run is idempotent.** A re-run for the same production day rewrites that\nday's report; it does not create a second one.\n\n---\n\n## Approval gates\n\n**This skill is read-only.** It reads, computes, prices and proposes, and it\nwrites nothing to any system of record — which is precisely why Chain 3 is safe\nto run unattended.\n\n**Asked to post, send or share the report** — to a Teams channel, by mail, to\nSlack — **say it does not post, and give the text ready to paste.** *\"I don't\npost to Teams. Here's the report as text to paste into the Production\nchannel.\"* No send tool is called, even where one is available, and **no offer\nto post later**: not \"confirm the channel and I will post this.\" The report is\nstill produced in full. Per `..\u002F..\u002Fshared\u002Fcommand-center.md`.\n\nTwo things are drafts requiring a human, not actions:\n\n- **The reschedule proposal** is drafted for the scheduler to accept or reject.\n- **A breakdown diagnosis carried in** stays labelled as advisory unless its\n  contract says a technician confirmed it.\n\n---\n\n## Output\n\n**The morning report is written to a file, and the reply is its summary.**\nSave the full page as `report-output\u002Fmorning-report.html`, one file saved over\nthe last run's, as the placement line says, with the site and production day\nin its title; name that path in the reply, which is the short summary in\n`..\u002F..\u002Fshared\u002Freplies.md` (about 150 words, 200 at most). A request that picks\nor skips sections is not the whole report: those sections go in the reply in\nfull. **With no way to write a file,\nprint the full page in the reply instead.** Any other document this skill\nproduces (a reschedule proposal handed to the scheduler, text to paste) is\nprinted in the reply in full.\n**A record handed to another skill is saved and summarised, never pasted as\nraw YAML**; memory blocks are written where the plant's memory lives; and the\nname on an approval is asked, never taken from the account. All three per\n`..\u002F..\u002Fshared\u002Freplies.md`.\n\n**The morning report** — an HTML page in the house style\n(`..\u002F..\u002Fshared\u002Fartifact-style.md`), and the Operations command-center page\n(`..\u002F..\u002Fshared\u002Fcommand-center.md`) — an artifact you place, never a write.\nChain 3 rewrites it overnight, which is the whole demonstration.\n\n**Run alone, this skill does not publish.** It never creates or updates the\nOperations Command Center dashboard and never writes a day record: only\n`morning-plant-pulse` does, per `..\u002F..\u002Fshared\u002Flive-dashboard.md`. A standalone\nrun returns its page in the conversation with the placement line, *\"save it\nover yesterday's,\"* and never says the dashboard was updated. Two dashboards\nshowing the same numbers is how a meeting ends up arguing about which is\nright. **The page leads with what the meeting acts on and keeps all the\nevidence below it**. In order:\n\n1. **Safety blocks, whenever they have anything in them.** A safety-flagged\n   diagnosis record first, then its own section headed exactly \"Blocked —\n   safety status unknown\", above the freshness banner\n2. **Freshness banner** — required, directly under any safety block and first\n   otherwise, never a footnote. It states the shifts the day actually holds;\n   a day with fewer shifts than usual says so, and is not a complete plant day\n3. **Headline figures**: good output against plan, total output beside good\n   when they differ, OEE with its three components, downtime minutes, and\n   scrap at standard cost (an estimate). Each caveat sits on its own figure\n4. **Today's actions**: one row per exception that needs someone, with the\n   issue, its impact (on a commitment where traceable), the next step, the\n   owner role, and its type. **A decision** is something genuinely waiting\n   on an approval: a normalization not yet approved, a reschedule move for the\n   scheduler. **An investigation** is something to find out first: an unknown\n   work order, a contradicted record, an uncoded stop. **An observation**\n   needs no action today. The owner role is the one the source names, else\n   *\"suggested: \u003Crole>\"*; never a person's name the data does not hold. With\n   nothing waiting, say *\"No decisions waiting today.\"* Never present a\n   reschedule, a reserved capacity or a root cause as already approved\n5. **Floor view**: the schematic, its legend, *\"State as of the last\n   export, [time]. Not live.\"* and the lists under it, per\n   `reference\u002Ffloor-schematic.md`. Every number is a figures-block row. With\n   no layout, one line\n6. **The detail**, every table the report has always held, collapsible but\n   never removed: output against plan with variances; OEE per asset or line,\n   three components separate; downtime ranked by minutes and by frequency;\n   priced losses with the rate assumption stated; week-over-week movers,\n   omitted entirely when there is no earlier day to compare; exceptions with\n   their commitment impact; reschedule proposals, one per significant\n   exception; then sources, methods and assumptions\n\n**Safety leads the report, freshness comes straight after it.** A safety\nimplication goes before anything else, and a banner of source ages is not\na reason for a lockout instruction to sit second. **When there is a safety\nblock, it is the first thing under the title.** The title names the site and\nthe production day, nothing else: no data age, no scope line, no intro\nsentence above the safety block. *\"Fairview: morning report, 31 Aug\"*, then\nthe safety block, then the freshness banner. Both stay above every\nranking and figure. \"Blocked — safety status unknown\" lists every diagnosis\nrecord held out of the rankings. **\"Checked, nothing wrong\"** follows the\nexceptions, holding the one-sided records that were looked at and are fine —\noutside every problem count.\n\n**The structured data behind the report**, so every number can be checked rather\nthan taken on trust. **It includes the figures block** — required on every\nrun, attended or not: each stated figure, its value, and the filter that\nproduced it, per `..\u002F..\u002Fshared\u002Fcomputed-figures.md`.\n\n---\n\n## Edge cases and failure modes\n\n- **Last night's export did not land** — report the gap prominently. Never\n  silently reuse yesterday's data.\n- **Partial data, one line's export missing** — report on what is there, name\n  what is missing. Never let a section quietly shrink.\n- **Reason codes inconsistent or heavily \"Other\"** — report the coverage problem.\n  That is a finding of its own, not a data-cleaning task.\n- **A machine ran unscheduled overtime** — availability can exceed 100%. Report\n  it; do not clamp it.\n- **Three shifts crossing midnight** — the production-day attribution rule is\n  configured per site, and the report states which convention is in use.\n- **No earlier day in the history or the exports** — skip week-over-week.\n- **An asset in the plan with no output data** — it is unreported, **not 0% OEE**.\n  Name it and exclude it from averages, saying so.\n- **No standard cost available** — report scrap in units and say it could not be\n  priced.\n- **Anything in the floor view** (an unreadable plan, a label that could be\n  two assets, a down machine, an asset added later, a busy drawing) — per\n  `reference\u002Ffloor-schematic.md`, *Failure modes*.\n\n---\n\n## Degradation\n\n**If Limble isn't included, or no CMMS connector is live** — minor. The report loses its \"maintenance work\norders in progress\" line. Replace it from the maintenance export where one\nexists, or omit the line **and say so** — never leave a silently empty section,\nwhich reads as \"no open maintenance\" rather than \"we could not see it.\"\n\nChain 3 is unaffected in any meaningful way: it is an overnight batch, so export\ncadence is already the right cadence. The freshness banner names maintenance as\nexport-sourced so nobody assumes it is live.\n\n**Without a planning export** — output against plan cannot be computed. Report\nactual output and say plan comparison was unavailable.\n\n**Without a downtime cost rate** — price scrap only, and state that downtime is\nunpriced.\n\n**Without a floor layout** — one line saying none is on file and how to set\none up. Nothing else changes. Without OEE thresholds, the defaults, labelled.\n\n---\n\n## Chain position\n\n- **Chain 3, Morning plant pulse (scheduled)** — step 2, from `data-consolidate`.\n  The only scheduled chain, read-only throughout, which is why it needs no\n  approval gate.\n- **Chain 4, Breakdown to next report** — step 2, from `breakdown-diagnosis`.\n  Step 7 folds its downtime record into the log and exceptions with no\n  transformation, per `..\u002F..\u002Fshared\u002Fcontracts\u002Fdowntime-record.md`.\n\n**Metric.** OEE visibility; supervisor hours; downtime and scrap cost\nvisibility — step 5 prices both on every run, not only when a breakdown fed\nin.\n\n---\n\n## Reference files\n\n| File | Load when |\n|---|---|\n| `reference\u002Foee.md` | Step 3 — the three components, the denominator traps, and the never-substitute-1.0 rule |\n| `reference\u002Freschedule-scope.md` | Step 9 — what a proposal must contain and where the scope stops |\n| `reference\u002Ffloor-schematic.md` | Step 9a — the setup turn, the drawing rules, the OEE thresholds |\n| `..\u002F..\u002Fshared\u002Fcontracts\u002Ffloor-layout.md` | Step 9a — the stored layout's shape and missing behaviors |\n| `..\u002F..\u002Fshared\u002Fcontracts\u002Fconsolidated-dataset.md` | Step 1 — the primary input's schema and its missing behaviors |\n| `..\u002F..\u002Fshared\u002Fcontracts\u002Fdowntime-record.md` | Step 7 — the shape a breakdown diagnosis arrives in |\n| `..\u002F..\u002Fshared\u002Fapproval-gates.md` | Step 2 — a match that rests on an unapproved normalization is provisional |\n| `..\u002F..\u002Fshared\u002Fcomputed-figures.md` | Step 10 — the figures block every stated number is quoted from |\n","# Daily Production Report\n\nEvery plant runs a morning production meeting, and somebody spends 60 to 90\nminutes before it assembling the same report by hand out of the same four or\nfive exports. This is that report.\n\n**Chain 3 runs this overnight, so the page exists before anyone asks for it.**\nThe run is built to be safe unattended — it asks nothing, it is idempotent, and\nit reports its own gaps. **The trigger is the host's, not the plugin's**: see\n`..\u002F..\u002Fshared\u002Fscheduling.md`. Never tell a user the report has been scheduled.\nThat is the clearest demonstration of the plugin's concept, and it\nis why the run must be idempotent — a re-run rewrites the same report for that\nproduction day rather than producing a second one.\n\n**Quick start**\n\n```\n→ Confirm freshness; if last night's export didn't land, lead with that\n→ Output against plan by line, cell, product, with variances\n→ OEE per asset or line, three components always shown separately\n→ Downtime ranked by minutes and by frequency, both\n→ Price scrap and downtime in dollars, with the rate assumption stated\n→ Week-over-week movers from history: the pulse's day records, or the exports\n→ Fold in any breakdown diagnoses\n→ Exceptions, each with its impact on a commitment\n→ One traced reschedule move per significant exception, drafted not applied\n→ Floor view: each machine's figures drawn where it sits; set up once from a\n  floor plan, confirmed by a person, saved on the plant setup page\n→ Page saved to report-output\u002F; the reply is a ~150-word summary naming it\n→ Run alone: never publishes. End with the placement line, \"save it over yesterday's\"\n```\n\n---\n\n## Inputs\n\n**First, find the plant setup:** from the task's own folder (never `cd` into the plugin), run\n`python3 -I \u003Cthis skill's folder>\u002F..\u002F..\u002Fshared\u002Fscripts\u002Fplant_setup.py find`, do its `setup_page_first` step, then read the profile at\n`profile_path` and any rate at `rates_path`, per `..\u002F..\u002Fshared\u002Freplies.md`.\n\n| Input | Source | Form |\n|---|---|---|\n| Consolidated dataset | `data-consolidate` | **Primary input.** `..\u002F..\u002Fshared\u002Fcontracts\u002Fconsolidated-dataset.md` |\n| New breakdown diagnoses | `breakdown-diagnosis` | `..\u002F..\u002Fshared\u002Fcontracts\u002Fdowntime-record.md` |\n| Production plan \u002F schedule | Planning export to the M365 folder | CSV \u002F XLSX |\n| Actual output, good and scrap | MES export to the M365 folder | CSV \u002F XLSX |\n| Downtime events with reason codes | MES or historian export | CSV |\n| Standard cost per part | The ERP. Infor M3 gives none (`..\u002F..\u002Fshared\u002Ferp-connectors.md`): the profile's cost read order applies | Connector |\n| Open work orders and due dates | The ERP | Connector |\n| Maintenance work orders in progress | CMMS | CMMS connector, else the maintenance export |\n| Stored day history, inside Chain 3 only | `morning-plant-pulse` | `..\u002F..\u002Fshared\u002Fcontracts\u002Fpulse-day-record.md` |\n| Floor layout (optional) | The `### Floor layout - \u003Csite>` section of the plant profile at `profile_path` (the plant setup page) | `..\u002F..\u002Fshared\u002Fcontracts\u002Ffloor-layout.md` |\n| Floor plan (optional, setup only) | A PDF or image the user attaches | Read once, in the setup turn of step 9a; never stored |\n| OEE thresholds (optional) | The plant profile's `OEE thresholds` line | Defaults apply, labelled not confirmed, when absent |\n\n---\n\n## Workflow\n\n### Step 1 — Confirm freshness\n\nConfirm freshness via `data-consolidate` — the data quality report and each\nrow's `as_of`. **If last night's export did not land, say so at the top of the\nreport**, directly under any safety block. Nothing about data age goes above a\nsafety-flagged record, not even a one-line note.\n\nA source whose timestamps run later than now is flagged `future_timestamp`:\nsay its dates can't be trusted and give it no age. A production report built on\nstale data is worse than no production report,\nbecause it gets believed and acted on. Never silently reuse yesterday's data.\nThe rule is `..\u002F..\u002Fshared\u002Fabsent-is-not-zero.md`.\n\n**On reading both the dataset and the raw exports.** The dataset is the primary\ninput — it carries the freshness picture and the classified key mismatches. The\nplan, output and downtime exports give rows at their native grain. Where the two\ndisagree on a count, **report both and name which is which**; a discrepancy is a\nfinding about the export, not a number to choose between. See *What this dataset\nis authoritative for* in `..\u002F..\u002Fshared\u002Fcontracts\u002Fconsolidated-dataset.md`.\n\n### Step 2 — Output against plan\n\nBy line, cell and product, with variances. Use the plant's own word for the\norganizational unit, from the `## Plant context` block.\n\n**Where plan and output spell a part differently, do not match them silently,\nand do not refuse to match them.** Run on raw exports, the plan says\n`PN-4471-B` and the MES says `4471B`. Dropping the row makes output against\nplan look broken; joining it without a word applies a fix nobody approved. So:\n\n1. Use any normalization already approved — in the consolidated dataset, or\n   on the profile's `### Approved normalizations` list — as it stands.\n2. For every other match that depends on a normalization, **show the variance\n   labelled provisional**, and beside it **each normalization it rests on, with\n   both spellings verbatim** — `4471B` → `PN-4471-B`, format drift.\n3. **Ask for a yes on each**, the same \"apply?\" gate `data-consolidate` uses.\n   Until then the rows stay provisional, in the table and in any total built\n   on them.\n\nAn unattended run cannot ask: it carries the same provisional table and lists\nthe normalizations as *proposed, awaiting approval*. Per\n`..\u002F..\u002Fshared\u002Fapproval-gates.md`.\n\n### Step 3 — Compute OEE\n\nPer asset or line, **with the three components always shown separately**. Read\n`reference\u002Foee.md` first — the denominators carry real traps, and a missing\ncomponent is never substituted with 1.0.\n\n**State the quality convention on every run**, as a line under the OEE table:\n*\"quality = good ÷ (good + scrap)\"*. **Only when the export has no rework\ncolumn** does the line continue: *\"no rework column, so first-pass yield could\nnot be applied.\"* Where the export does carry rework, quality is first-pass\nyield, good ÷ (good + scrap + rework), and the line says so instead. Never\nprint the no-rework caveat against an export that has the column.\n\n### Step 4 — Build the downtime log\n\nRanked **two ways, both always shown**: by total minutes, and by frequency.\n\n**`Other` never appears in a cause ranking.** Before ranking by reason code,\ncount the events and minutes coded `Other` or left blank and state that share\nas a finding of its own: *\"18 of 39 events, 46%, carry no usable reason code.\"*\nThen rank the named codes only, with the caveat that the picture is incomplete\nby that share. A table with `Other` in row 1, priced, tells the meeting the\nplant's biggest problem is a word, and it is the single most common way this\nreport has been wrong.\n\n**A downtime row that ends before it starts, or whose duration disagrees with\nits own start and end, has no trustworthy minutes.** It carries\n`end_before_start` or `duration_mismatch` from the dataset; reading the export\nraw, test it in code. Hold its minutes out of every total and the minutes\nranking, count it as an event, and state the exclusion beside the ranking with\nthe row: *\"DOOSAN-PUMA-07, 4 Aug: ends 14:01 after starting 14:05, 56 minutes\nstated, not counted.\"* Never pick a figure for it.\n\nThese describe different problems. One long outage and forty chronic small\nstoppages can total the same minutes, and the second is usually the more fixable\nof the two. A single ranking hides whichever one it does not sort by.\n\n### Step 5 — Price the losses\n\nScrap × standard cost. Downtime × the plant's configured cost-per-hour.\n\n**Scrap dollars are an estimate at standard cost, and say so** where they are\nprinted (*\"$180 scrap, estimated at standard cost\"*): they are exposure, not\na booked loss. With no rate, downtime dollars are *unknown*, never $0.\n\n\"47 parts scrapped\" does not land in a production meeting. \"$8,400 of scrap, 71%\nof it on one part number\" does, and it is the same data multiplied by a cost.\n\n**Every quantity in the priced-losses table is the figures-block row, printed.**\nThe scrap count beside a part's dollar figure is the same number the dollar\nfigure was multiplied from; a table that says 155 units against a figure\ncomputed from 62 has been typed, not printed. Build the table from the block,\nper `..\u002F..\u002Fshared\u002Fcomputed-figures.md`, and never from memory of it.\n\n**A part whose scrap is recorded in a different unit from its standard cost is\nunpriced, not converted.** Scrap in `box12` against a cost per `each` has no\ndollar figure until the plant says what a box holds: report the quantity in\nits own unit, mark the dollar cell \"unpriced — unit conflict,\" and name the\nconflict. Never price it two ways (\"$5,465 if singles, $65,584 if boxes of\n12\"): both numbers are guesses, and the larger one is the one that gets\nrepeated. The rule is the same one `data-consolidate` applies at the join.\n\n**Always state the assumption behind the downtime rate.** Resolve it per\n`..\u002F..\u002Fshared\u002Fsecret-config.md`, *Resolving a rate* —\n`MFG_DOWNTIME_COST_PER_HOUR` if set, then a confirmed rate in the plant\nprofile, then a placeholder there, labelled \"placeholder rate — not\nconfirmed\" — and name the source next to the priced downtime. This is an\nunattended overnight run, so there's no one to ask. Without a rate in either\nroute, price scrap only and say plainly that downtime is unpriced, per\n`..\u002Fonboarding\u002Freference\u002Fconfig-profile.md`.\n\n### Step 6 — Week-over-week movers\n\nCompare against earlier production days and surface the biggest improvements\nand deteriorations. **The history comes from one of two places, never from a\nprior page.** Inside Chain 3, `morning-plant-pulse` hands over the stored day\nrecords. Run alone, this skill compares only against earlier days the exports\nin front of it cover, and says that is the basis. Trend beats snapshot: a 68% OEE that was 61% last week is a\ndifferent conversation from one that was 74%.\n\n**Compare all four: output against plan, OEE by component, downtime minutes and\nscrap dollars.** A movers section that shows only OEE has missed the number the\nplant manager looks at first. Each mover names the metric, last week's value,\nthis week's, and the change.\n\n**With no earlier day to compare against, skip the section rather than\nshowing an empty comparison.**\n\n### Step 7 — Fold in breakdown diagnoses\n\nAny downtime record from `breakdown-diagnosis` appends into the downtime log and\nthe exceptions, with no transformation step. Its contract governs how — including\nthat an ongoing outage has a null duration rather than zero, that a safety flag\nleads with its `safety_reason`, and that an advisory cause is attributed as\nproposed rather than established, with its `technician_step` beside it\n(*\"proposed; technician step skipped\"*).\n\n**Check `safety_flagged` on every record where `source_system` is\n`breakdown_diagnosis`, before it goes anywhere:**\n\n- **Absent — block the record.** It appears only in a separate **\"Blocked —\n  safety status unknown\"** list, **placed first, above both rankings**, with\n  its asset, start time and the reason: *safety status missing on a diagnosis\n  record — not ranked until confirmed.* It is not in the minutes ranking, not in the frequency ranking, not in any\n  total, and not priced, until the flag is known. Labelling it \"unknown\" and\n  ranking it anyway is not blocking it. Never read a missing flag as false.\n- **True — it leads the report**, above both rankings.\n- **False** — it joins the log as an ordinary row.\n\n**No diagnosis records at all is not a clear safety status.** When the drop\nholds none, write *\"Safety status: not assessed (no breakdown diagnosis records\nin this drop)\"* once, where the safety block would sit. Never \"no safety\nissues\", and never read an empty maintenance or quality export as nothing open.\n\n**Rows from the MES or historian export are not contract records and are left\nexactly as they are.** They never carry this field; its absence there means\nnothing, and blocking them would empty the report.\n\n### Step 8 — Identify exceptions\n\nLate jobs, machines down, material shortages, quality holds. **Each with its\nimpact on a commitment** — an exception with no stated consequence is not\nactionable. The commitment is read through the work order's `sales_order`.\nWhere the source holds no work order to sales order link, say *\"commitment\nimpact not traceable: no work order to sales order link in the ERP\"* once,\nwith the count it affects, and never infer the order from part and customer.\n\n**A genuine one-sided record is not an exception, and is never counted as\none.** A part in the item master with no plan and no production, a machine\nwith no downtime, a supplier with no purchase orders in the window: that is\nClass 6 in `..\u002Fdata-consolidate\u002Freference\u002Fmismatch-classes.md` — *report, do\nnot alarm* — and usually nothing is wrong at all. Build the list in two\npasses:\n\n1. Classify each candidate first. Class 6 records leave the exceptions list\n   and every count of exceptions or problems.\n2. Show them in a separate **\"Checked, nothing wrong\"** block, one line each:\n   *\"in the item master, not planned and not produced this month.\"*\n\nDecide it by class, from the record's own type and dates, never by recognising\na part number. **Route by class number before numbering any row**: classes 1,\n2, 3, 4, 5 and 8 are exceptions or problems; class 7 is the coverage line in\nthe freshness banner; class 6 is \"Checked, nothing wrong\" **unless the record\ncarries `contradicted_by`**. A one-sided record that another source\ncontradicts (WO-88412, closed with no output, while NCR-2026-0141 found bad\nparts from it) is an exception, *\"one-sided, contradicted by NCR-2026-0141\"*,\nand never sits under \"nothing wrong\". A duplicate key (class 8) is counted by\nthe rows that collide at the tab's grain, from the query: WO-88201 has three\nrows, two of them (509 and 480, both 07:00) collide, and the 95 on the 23:00\nshift is its own shift. State both numbers. One false alarm in the list is what makes a supervisor stop\nreading the rest of it. A date-window gap (Class 7) is different: it stays a\ncoverage statement in the freshness banner, and is not a record-level\nexception either.\n\n### Step 9 — Propose the reschedule or changeover move\n\nFor each significant exception: the specific move, what it displaces, what still\nmisses its date, and what was checked versus not checked.\n\n**Read `reference\u002Freschedule-scope.md` before doing this.** There is a hard scope\nline — one or two traced moves, never a re-optimised day — and the reason is that\na plausible-looking schedule that quietly violates constraints is worse than no\nproposal at all.\n\nDrafted for the scheduler. Never applied.\n\n### Step 9a — Floor view\n\n**Read `reference\u002Ffloor-schematic.md` first.** It holds the setup turn, the\nsave, the drawing rules, the states, the thresholds and the failure modes.\n\n- **The user attaches a floor plan, or asks for a floor view and none is on\n  file:** run the setup turn. State the count first, propose each match with\n  both spellings verbatim, list what did not match on either side, and **ask\n  for a yes per area**. Turn 1 shows, turn 2 saves, and only with a name or\n  role the person typed. Per `..\u002F..\u002Fshared\u002Fapproval-gates.md`.\n- **A layout is on file** (the `### Floor layout - \u003Csite>` section of the\n  profile): draw it from the figures block. **Compute nothing new.** An\n  unconfirmed mapping is a neutral outline with no figure. No data is grey,\n  never 0%. Data but no computable OEE is grey *OEE N\u002FA* with its reason.\n- **No layout:** one line saying none is on file and how to set one up.\n\n**An unattended run never sets up, edits or re-maps a layout**, and never\ndraws a guess. Colours follow the profile's `OEE thresholds` line; until the\nplant confirms one, the defaults are used and the legend says so.\n\n### Step 10 — Emit\n\n**Build the figures block first, in code, then write from it.** Every count,\ntotal, percentage and \"N of M\" in the report — events per reason code, events\nsharing a timestamp, minutes per asset, scrap share per part — is printed from\na computation run in this session and listed in the block with its filter. The\nprose and the tables quote the block; nothing is retyped from memory. Per\n`..\u002F..\u002Fshared\u002Fcomputed-figures.md`. \"Thirty of the 39 events\" written beside a\nlog that holds thirty-one is what this step exists to stop.\n\nRender the report and the structured data behind it, per Output below. Chain 3\nemits overnight so the report exists before the morning meeting.\n\n**The run is idempotent.** A re-run for the same production day rewrites that\nday's report; it does not create a second one.\n\n---\n\n## Approval gates\n\n**This skill is read-only.** It reads, computes, prices and proposes, and it\nwrites nothing to any system of record — which is precisely why Chain 3 is safe\nto run unattended.\n\n**Asked to post, send or share the report** — to a Teams channel, by mail, to\nSlack — **say it does not post, and give the text ready to paste.** *\"I don't\npost to Teams. Here's the report as text to paste into the Production\nchannel.\"* No send tool is called, even where one is available, and **no offer\nto post later**: not \"confirm the channel and I will post this.\" The report is\nstill produced in full. Per `..\u002F..\u002Fshared\u002Fcommand-center.md`.\n\nTwo things are drafts requiring a human, not actions:\n\n- **The reschedule proposal** is drafted for the scheduler to accept or reject.\n- **A breakdown diagnosis carried in** stays labelled as advisory unless its\n  contract says a technician confirmed it.\n\n---\n\n## Output\n\n**The morning report is written to a file, and the reply is its summary.**\nSave the full page as `report-output\u002Fmorning-report.html`, one file saved over\nthe last run's, as the placement line says, with the site and production day\nin its title; name that path in the reply, which is the short summary in\n`..\u002F..\u002Fshared\u002Freplies.md` (about 150 words, 200 at most). A request that picks\nor skips sections is not the whole report: those sections go in the reply in\nfull. **With no way to write a file,\nprint the full page in the reply instead.** Any other document this skill\nproduces (a reschedule proposal handed to the scheduler, text to paste) is\nprinted in the reply in full.\n**A record handed to another skill is saved and summarised, never pasted as\nraw YAML**; memory blocks are written where the plant's memory lives; and the\nname on an approval is asked, never taken from the account. All three per\n`..\u002F..\u002Fshared\u002Freplies.md`.\n\n**The morning report** — an HTML page in the house style\n(`..\u002F..\u002Fshared\u002Fartifact-style.md`), and the Operations command-center page\n(`..\u002F..\u002Fshared\u002Fcommand-center.md`) — an artifact you place, never a write.\nChain 3 rewrites it overnight, which is the whole demonstration.\n\n**Run alone, this skill does not publish.** It never creates or updates the\nOperations Command Center dashboard and never writes a day record: only\n`morning-plant-pulse` does, per `..\u002F..\u002Fshared\u002Flive-dashboard.md`. A standalone\nrun returns its page in the conversation with the placement line, *\"save it\nover yesterday's,\"* and never says the dashboard was updated. Two dashboards\nshowing the same numbers is how a meeting ends up arguing about which is\nright. **The page leads with what the meeting acts on and keeps all the\nevidence below it**. In order:\n\n1. **Safety blocks, whenever they have anything in them.** A safety-flagged\n   diagnosis record first, then its own section headed exactly \"Blocked —\n   safety status unknown\", above the freshness banner\n2. **Freshness banner** — required, directly under any safety block and first\n   otherwise, never a footnote. It states the shifts the day actually holds;\n   a day with fewer shifts than usual says so, and is not a complete plant day\n3. **Headline figures**: good output against plan, total output beside good\n   when they differ, OEE with its three components, downtime minutes, and\n   scrap at standard cost (an estimate). Each caveat sits on its own figure\n4. **Today's actions**: one row per exception that needs someone, with the\n   issue, its impact (on a commitment where traceable), the next step, the\n   owner role, and its type. **A decision** is something genuinely waiting\n   on an approval: a normalization not yet approved, a reschedule move for the\n   scheduler. **An investigation** is something to find out first: an unknown\n   work order, a contradicted record, an uncoded stop. **An observation**\n   needs no action today. The owner role is the one the source names, else\n   *\"suggested: \u003Crole>\"*; never a person's name the data does not hold. With\n   nothing waiting, say *\"No decisions waiting today.\"* Never present a\n   reschedule, a reserved capacity or a root cause as already approved\n5. **Floor view**: the schematic, its legend, *\"State as of the last\n   export, [time]. Not live.\"* and the lists under it, per\n   `reference\u002Ffloor-schematic.md`. Every number is a figures-block row. With\n   no layout, one line\n6. **The detail**, every table the report has always held, collapsible but\n   never removed: output against plan with variances; OEE per asset or line,\n   three components separate; downtime ranked by minutes and by frequency;\n   priced losses with the rate assumption stated; week-over-week movers,\n   omitted entirely when there is no earlier day to compare; exceptions with\n   their commitment impact; reschedule proposals, one per significant\n   exception; then sources, methods and assumptions\n\n**Safety leads the report, freshness comes straight after it.** A safety\nimplication goes before anything else, and a banner of source ages is not\na reason for a lockout instruction to sit second. **When there is a safety\nblock, it is the first thing under the title.** The title names the site and\nthe production day, nothing else: no data age, no scope line, no intro\nsentence above the safety block. *\"Fairview: morning report, 31 Aug\"*, then\nthe safety block, then the freshness banner. Both stay above every\nranking and figure. \"Blocked — safety status unknown\" lists every diagnosis\nrecord held out of the rankings. **\"Checked, nothing wrong\"** follows the\nexceptions, holding the one-sided records that were looked at and are fine —\noutside every problem count.\n\n**The structured data behind the report**, so every number can be checked rather\nthan taken on trust. **It includes the figures block** — required on every\nrun, attended or not: each stated figure, its value, and the filter that\nproduced it, per `..\u002F..\u002Fshared\u002Fcomputed-figures.md`.\n\n---\n\n## Edge cases and failure modes\n\n- **Last night's export did not land** — report the gap prominently. Never\n  silently reuse yesterday's data.\n- **Partial data, one line's export missing** — report on what is there, name\n  what is missing. Never let a section quietly shrink.\n- **Reason codes inconsistent or heavily \"Other\"** — report the coverage problem.\n  That is a finding of its own, not a data-cleaning task.\n- **A machine ran unscheduled overtime** — availability can exceed 100%. Report\n  it; do not clamp it.\n- **Three shifts crossing midnight** — the production-day attribution rule is\n  configured per site, and the report states which convention is in use.\n- **No earlier day in the history or the exports** — skip week-over-week.\n- **An asset in the plan with no output data** — it is unreported, **not 0% OEE**.\n  Name it and exclude it from averages, saying so.\n- **No standard cost available** — report scrap in units and say it could not be\n  priced.\n- **Anything in the floor view** (an unreadable plan, a label that could be\n  two assets, a down machine, an asset added later, a busy drawing) — per\n  `reference\u002Ffloor-schematic.md`, *Failure modes*.\n\n---\n\n## Degradation\n\n**If Limble isn't included, or no CMMS connector is live** — minor. The report loses its \"maintenance work\norders in progress\" line. Replace it from the maintenance export where one\nexists, or omit the line **and say so** — never leave a silently empty section,\nwhich reads as \"no open maintenance\" rather than \"we could not see it.\"\n\nChain 3 is unaffected in any meaningful way: it is an overnight batch, so export\ncadence is already the right cadence. The freshness banner names maintenance as\nexport-sourced so nobody assumes it is live.\n\n**Without a planning export** — output against plan cannot be computed. Report\nactual output and say plan comparison was unavailable.\n\n**Without a downtime cost rate** — price scrap only, and state that downtime is\nunpriced.\n\n**Without a floor layout** — one line saying none is on file and how to set\none up. Nothing else changes. Without OEE thresholds, the defaults, labelled.\n\n---\n\n## Chain position\n\n- **Chain 3, Morning plant pulse (scheduled)** — step 2, from `data-consolidate`.\n  The only scheduled chain, read-only throughout, which is why it needs no\n  approval gate.\n- **Chain 4, Breakdown to next report** — step 2, from `breakdown-diagnosis`.\n  Step 7 folds its downtime record into the log and exceptions with no\n  transformation, per `..\u002F..\u002Fshared\u002Fcontracts\u002Fdowntime-record.md`.\n\n**Metric.** OEE visibility; supervisor hours; downtime and scrap cost\nvisibility — step 5 prices both on every run, not only when a breakdown fed\nin.\n\n---\n\n## Reference files\n\n| File | Load when |\n|---|---|\n| `reference\u002Foee.md` | Step 3 — the three components, the denominator traps, and the never-substitute-1.0 rule |\n| `reference\u002Freschedule-scope.md` | Step 9 — what a proposal must contain and where the scope stops |\n| `reference\u002Ffloor-schematic.md` | Step 9a — the setup turn, the drawing rules, the OEE thresholds |\n| `..\u002F..\u002Fshared\u002Fcontracts\u002Ffloor-layout.md` | Step 9a — the stored layout's shape and missing behaviors |\n| `..\u002F..\u002Fshared\u002Fcontracts\u002Fconsolidated-dataset.md` | Step 1 — the primary input's schema and its missing behaviors |\n| `..\u002F..\u002Fshared\u002Fcontracts\u002Fdowntime-record.md` | Step 7 — the shape a breakdown diagnosis arrives in |\n| `..\u002F..\u002Fshared\u002Fapproval-gates.md` | Step 2 — a match that rests on an unapproved normalization is provisional |\n| `..\u002F..\u002Fshared\u002Fcomputed-figures.md` | Step 10 — the figures block every stated number is quoted from |\n",{"name":5,"description":7,"compatibility":24},"Requires a consolidated dataset from data-consolidate. Reads the ERP (Oracle NetSuite or Infor M3) for standard cost and open work orders, and a CMMS connector (Limble is the worked example) for maintenance work orders in progress, when connected. Runs on export-only data. Writes nothing to any system of record.","main","10759d7fff7d18eecf689010d3fc2543b55bb7a9","plugins\u002Fclaude-for-manufacturing\u002Fskills\u002Fdaily-production-report",4913,"https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fclaude-plugins-official","https:\u002F\u002Fcode.claude.com\u002Fdocs\u002Fen\u002Fplugins","https:\u002F\u002Fgithub.com\u002Fanthropics","https:\u002F\u002Fanthropic.com","2026-11-14 15:27:28",{"source":8,"skillName":5,"commands":35,"prompt":45,"zipAvailable":46},[36,39,42],{"tool":37,"command":38},"npx","npx skills add anthropics\u002Fclaude-plugins-official --skill daily-production-report",{"tool":40,"command":41},"bunx","bunx skills add anthropics\u002Fclaude-plugins-official --skill daily-production-report",{"tool":43,"command":44},"pnpm","pnpm dlx skills add anthropics\u002Fclaude-plugins-official --skill daily-production-report","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fclaude-plugins-official\u002Fblob\u002Fmain\u002Fplugins\u002Fclaude-for-manufacturing\u002Fskills\u002Fdaily-production-report\u002FSKILL.md",true,[48,54,58],{"path":49,"resourceType":50,"resourceTypeDesc":51,"sizeBytes":52,"sourceHtmlUrl":53},"reference\u002Ffloor-schematic.md","REFERENCE","参考文档",13452,"https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fclaude-plugins-official\u002Fblob\u002Fmain\u002Fplugins\u002Fclaude-for-manufacturing\u002Fskills\u002Fdaily-production-report\u002Freference\u002Ffloor-schematic.md",{"path":55,"resourceType":50,"resourceTypeDesc":51,"sizeBytes":56,"sourceHtmlUrl":57},"reference\u002Foee.md",5965,"https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fclaude-plugins-official\u002Fblob\u002Fmain\u002Fplugins\u002Fclaude-for-manufacturing\u002Fskills\u002Fdaily-production-report\u002Freference\u002Foee.md",{"path":59,"resourceType":50,"resourceTypeDesc":51,"sizeBytes":60,"sourceHtmlUrl":61},"reference\u002Freschedule-scope.md",4412,"https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fclaude-plugins-official\u002Fblob\u002Fmain\u002Fplugins\u002Fclaude-for-manufacturing\u002Fskills\u002Fdaily-production-report\u002Freference\u002Freschedule-scope.md",[63,70,77,84,91,98],{"skillNo":64,"name":65,"nameZh":6,"description":66,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":67,"categories":68,"keywords":69,"publishedAt":19},"01M78Q264BAEFG71WSTBJZJWXN","trade-study-report","Turns a filled-in trade study matrix and its analysis results into a scored comparison and a decision memo, with every score cited to its source. Runs the sensitivity analysis every time — where the winner flips, which criterion the decision actually rests on, whether an option never wins under any plausible weighting — because a recommendation with no sensitivity check is a spreadsheet, not a decision. Disqualifies options that violate a hard constraint rather than scoring them low. Consumes design-space-explore's variant manifest when one exists, or a matrix filled in by hand when it doesn't. Use when someone says \"compare these options,\" \"write the trade study,\" \"which design should we pick,\" \"score these against the requirements,\" or \"how sensitive is this decision to the weighting.\"","https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fclaude-plugins-official\u002Fblob\u002Fmain\u002Fplugins\u002Fclaude-for-manufacturing\u002Fskills\u002Ftrade-study-report\u002FSKILL.md",[],[],{"skillNo":71,"name":72,"nameZh":6,"description":73,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":74,"categories":75,"keywords":76,"publishedAt":19},"01M78Q1VQ9XH9A0VGGH217E17Z","supplier-quote-request","Builds the RFQ package that goes out to suppliers — drawing and specification at the verified current revision, the demand picture, the quality clause list, commercial terms, and a mandated response format that makes the returned bids comparable by construction. Lists everything missing. Hard stops on a revision mismatch or restricted technical data. Then runs the supplier loop: candidates from the approved supplier list with unapproved and single-source flagged, one email draft per supplier for the buyer to send, response tracking with chase drafts, returned bids read into one shape with every missing field listed, a comparison per quantity break computed in code, and a recommendation. The buyer decides; the skill records who and when. Use when someone says \"source this part,\" \"send an RFQ to suppliers,\" \"we need an alternate supplier,\" \"get quotes on this,\" \"who else can make this?\", \"chase the suppliers,\" \"compare these bids,\" or \"we're going with this supplier.\"","https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fclaude-plugins-official\u002Fblob\u002Fmain\u002Fplugins\u002Fclaude-for-manufacturing\u002Fskills\u002Fsupplier-quote-request\u002FSKILL.md",[],[],{"skillNo":78,"name":79,"nameZh":6,"description":80,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":81,"categories":82,"keywords":83,"publishedAt":19},"01M78Q1MTXB0PTT396QS4NQMNR","supplier-escape-to-alternate-source","Runs the cross-area chain from a nonconformance traced to purchased material through to an alternate-source RFQ: writes the NCR and its containment, establishes root cause against the supplier rather than the process, records the corrective action, then builds the sourcing package with a response format that makes the returned bids comparable. Use when someone says \"run the supplier escape chain,\" \"a supplier caused this, find us an alternate,\" \"supplier escape to alternate source,\" or asks to source around a supplier who caused a defect.","https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fclaude-plugins-official\u002Fblob\u002Fmain\u002Fplugins\u002Fclaude-for-manufacturing\u002Fskills\u002Fsupplier-escape-to-alternate-source\u002FSKILL.md",[],[],{"skillNo":85,"name":86,"nameZh":6,"description":87,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":88,"categories":89,"keywords":90,"publishedAt":19},"01M78Q1C9YCXG5WWVZ4M5BQDM0","root-cause-capa","Reads sensor history, inspection measurements, maintenance records and prior tickets onto one verified timeline, runs the pattern checks in code rather than by eye — SPC drift, clustering by lot, shift, operator, machine and tool change, correlation with a maintenance event — then proposes ranked candidate causal chains, each with the evidence for it, the evidence against it, and the test that would confirm or kill it. Drafts the 8D. Where the leading candidate points at purchased material, raises the supplier-caused flag that starts a sourcing action. Use when someone says \"why did this happen,\" \"root cause,\" \"write the 8D,\" \"investigate this nonconformance,\" \"CAPA,\" or \"the customer wants a corrective action report.\"","https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fclaude-plugins-official\u002Fblob\u002Fmain\u002Fplugins\u002Fclaude-for-manufacturing\u002Fskills\u002Froot-cause-capa\u002FSKILL.md",[],[],{"skillNo":92,"name":93,"nameZh":6,"description":94,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":95,"categories":96,"keywords":97,"publishedAt":19},"01M78Q13PMV5X90HCCCFXF9FQE","rfq-to-quote","Runs the whole RFQ-to-quote chain end to end: reads the inquiry and its drawings, checks it against what the plant can build, drafts the questions back, returns a provisional bid\u002Fno-bid, pulls cost history, routings and closed-job actuals, builds an internal estimating model in Fusion where connected (from the drawing, a prior part's model saved as new, text specs as a ROM only, or a STEP file), derives machine hours per operation with each hour's basis labelled, and drafts the quote with its cost basis shown. Four approval gates, and the chain pauses for days while the customer answers. Use when someone says \"run the RFQ chain,\" \"take this RFQ all the way to a quote,\" \"quote this end to end,\" \"quote this from the drawing,\" or names an inquiry and asks for the full workflow.","https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fclaude-plugins-official\u002Fblob\u002Fmain\u002Fplugins\u002Fclaude-for-manufacturing\u002Fskills\u002Frfq-to-quote\u002FSKILL.md",[],[],{"skillNo":99,"name":100,"nameZh":6,"description":101,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":102,"categories":103,"keywords":104,"publishedAt":19},"01M78Q0W2VXEVHQDX24Q323040","quote-build","Produces the make-to-order quote from the intake form, comparable prior jobs and cost history, with every number traced to where it came from — material at the price actually paid on a named purchase order, machine hours per operation from a labelled ladder (current routing, closed-job actuals, the confirmed design, first principles) with setup and run kept apart, setup re-amortized at each quantity break, and margin on its own line. Checks the promised date against real shop load before committing to it. Revises in place when a customer counters, preserving what changed and why. Use when someone says \"build the quote,\" \"price this job,\" \"what should we charge,\" \"quote this RFQ,\" or \"the customer came back on price.\"","https:\u002F\u002Fgithub.com\u002Fanthropics\u002Fclaude-plugins-official\u002Fblob\u002Fmain\u002Fplugins\u002Fclaude-for-manufacturing\u002Fskills\u002Fquote-build\u002FSKILL.md",[],[]]