← Samaya Finance SkillsCONTENT HISTORY

Update to Samaya Finance Skills

Snapshot Nov 16, 2026 · 18:17 UTC · version 1.2.2

Collection source: downloaded plugin package.

WHAT CHANGED

First saved snapshot

No earlier snapshot is available to establish a change.

Save Samaya Finance Skills and check future changes in Following.

Up to 3 plugins free · no card required.

Compare saved observations

Download comparison JSON
Full technical diff · 0 changed fields
Full snapshot data
{
  "description": "\"earnings recap\", \"what actually changed on the call\" — the second read on a results event: the biggest positive and negative surprises in management's commentary against the release and the prior call, the key numbers against consensus, the debates the buy side pressed ranked by how well management answered them, and the next dated catalyst. Built around the two to four things the market actually judges this company on. Also triggers on \"debrief the call\", \"earnings call summary\", \"what did they say on the call\", \"post-earnings note\", \"recap the print\", \"how did management handle the Q&A\". For the variance grid alone use samaya-variance-table.\n",
  "included_files": [
    {
      "relative_path": "references/document-architecture.md",
      "size_in_bytes": 6484
    },
    {
      "relative_path": "references/hard-rules.md",
      "size_in_bytes": 3931
    },
    {
      "relative_path": "references/samaya-conventions.md",
      "size_in_bytes": 17837
    },
    {
      "relative_path": "references/samaya-house-language.md",
      "size_in_bytes": 8409
    },
    {
      "relative_path": "references/section-election.md",
      "size_in_bytes": 8104
    },
    {
      "relative_path": "references/template-and-sections.md",
      "size_in_bytes": 14615
    },
    {
      "relative_path": "references/validation-gates.md",
      "size_in_bytes": 6451
    },
    {
      "relative_path": "references/writing-standards.md",
      "size_in_bytes": 8542
    }
  ],
  "name": "samaya-earnings-recap",
  "skill_md_contents": "---\nname: samaya-earnings-recap\ndescription: >\n  \"earnings recap\", \"what actually changed on the call\" — the second read on a results event:\n  the biggest positive and negative surprises in management's commentary against the release and\n  the prior call, the key numbers against consensus, the debates the buy side pressed ranked by\n  how well management answered them, and the next dated catalyst. Built around the two to four\n  things the market actually judges this company on. Also triggers on \"debrief the call\",\n  \"earnings call summary\", \"what did they say on the call\", \"post-earnings note\", \"recap the\n  print\", \"how did management handle the Q&A\". For the variance grid alone use\n  samaya-variance-table.\n---\n\n# Earnings Call Debrief\n\nShared rules — the connectivity gate, consensus, citation, traceability, the attention fill,\ndesk, authorship and the common validation gates — live in `references/samaya-conventions.md`,\nand the page-one layout in `references/document-architecture.md`, and the writing standard — how every line is worded — in\n`references/samaya-house-language.md`. Read all three before the first retrieval; each section\nbelow keeps its non-negotiables and points there for the full rule.\n\nThis debriefs the results event as a whole — the release and the call together.\nThe release carries the numbers; the call carries the reasoning, the hedging, the\npushback and the things management chose not to say. The release is reported in its\nown right, and the call's contribution is stated as the delta against what the print\nalready told you — with anything drawn from the release labelled as such, so the\nreader can see which findings needed the call.\n\nThe output is a debrief for someone who could not attend, or who attended and\nwants the second read. It is not a transcript summary, and it never passes a figure\nalready in the release off as something the call revealed.\n\n**The deliverable is the differentiated finding, not coverage of the event.** The reader can get\na summary anywhere. What they cannot get anywhere is the one or two things said on this call\nthat change how the business should be understood — a diagnosis restated from market-driven to\nself-inflicted, a price position conceded, a reinvestment implied but not quantified. Those are\nthe report. Everything else is context for them, and context belongs behind them.\n\n**Language is evidence, not colour.** A beat can trade down because the tone turned, and a miss\ncan trade up because the explanation narrowed. Treat what management said, and how the saying\nchanged since last quarter, with the same seriousness as what they printed — and put both in\nfront of the reader in the same form, as tables they can scan rather than paragraphs they must\nmine.\n\n**Length is a cost, not a virtue.** This report earns its place by being read in full on the way\ninto a morning meeting. See the length budget below: it is a hard constraint, not an aspiration,\nand a section that cannot justify its space is cut rather than shortened.\n\n## STEP 0 — SAMAYA CONNECTIVITY GATE (hard requirement, before anything else)\n\nRead `references/samaya-conventions.md` first — it carries the full rule for this and for every\nshared section below. Then: confirm `financial_qa` and `financial_qa_result` are in the tool list\n(match the tool name, not the client's prefix; if absent, stop and ask the analyst to connect\nSamaya); fire one trivial test query and get a real answer before building; on failure retry the\nsame session, then a fresh one, then report and pause — never a silent fallback to web search or\nany other source; any non-Samaya source needs the analyst's explicit approval, per source, and is\nlabelled with its true retrieval path. An uploaded transcript alone is a build source only where the user elected the upload path in Step 0b.\n\n## STEP 0b — SUBJECT AND SOURCE\n\nResolve in one short turn. Do not proceed on a guess for the first two.\n\n1. **Company and period.** Resolve the ticker to its legal name via Samaya.\n   Confirm which call: the quarter or half, and the date it was held.\n2. **Transcript source.** The call transcript retrieved through Samaya, or a\n   transcript the user uploads. Where the retrieved transcript is partial,\n   say which portions are missing before building anything on them.\n3. **Consensus provider — blocking, and never defaulted.** Ask for the provider and\n   the vintage date before anything is retrieved: Visible Alpha, a company-compiled\n   consensus from the investor relations page, another mapped feed, or a file the\n   analyst supplies. **Do not default to any provider.** Two runs on the same company\n   days apart have landed on different providers because nobody chose, and the two\n   reports then disagreed about whether the print missed, on the same reported number.\n   Where the company itself points to a compiled consensus, say so and offer it, since\n   that is usually the one the market is quoting.\n   **The provider and its vintage date are printed in the consensus column header and\n   in the event strip**, so a reader sees what the comparison is against without\n   opening the appendix. Never blend providers, within a line or across the document.\n4. **Comparison baseline.** Tone, certainty language and the key metrics are read\n   against a trailing baseline — the reported period and the prior three comparable\n   periods. Confirm the trailing set; default to the three immediately preceding\n   comparable calls and periods.\n\n## STEP 0c — WHAT THIS COMPANY IS JUDGED ON — established before anything is built\n\nFull section: `references/section-election.md` — read it at kickoff, before presenting the menu.\n\n## STEP 0d — SECTION ELECTION — run the defaults, then say what else exists\n\nWhere the invocation resolves a company, **run the default sections\nimmediately.** Do not hold the run to present a menu.\n\nFull section: `references/section-election.md` — read it at kickoff, before presenting the menu.\n\n## HEADINGS AND LANGUAGE — written the way an analyst writes, not the way a model writes\n\nFull section: `references/writing-standards.md` — read it before writing the first section.\n\n## THE TEMPLATE — fixed, and the same for every company and every sector\n\nFull section: `references/template-and-sections.md` — read it before building the document.\n\n## THE VERDICT — the block at the top, on every run\n\nFull section: `references/template-and-sections.md` — read it before building the document.\n\n## 01 — WHAT CHANGED ON THE CALL, AND WHEN IT GETS TESTED\n\nOne table, carrying what changed, why it matters and the date it gets tested. Columns:\n\nFull section: `references/template-and-sections.md` — read it before building the document.\n\n## 02 — KEY NUMBERS\n\nFull section: `references/template-and-sections.md` — read it before building the document.\n\n## 03 — SEGMENT AND BRAND PERFORMANCE\n\nTwo charts, each with a takeaway line above it and a table underneath carrying the same figures.\nCharts are drawn from the retrieved figures, never from a separately recalled series.\n\nFull section: `references/template-and-sections.md` — read it before building the document.\n\n## 04 — KEY DEBATES FROM THE EARNINGS CALL\n\nColumns: the debate, the brokers who raised it and how it was handled, and what was actually\ngiven.\n\nFull section: `references/template-and-sections.md` — read it before building the document.\n\n## 05 — CHANGE IN MANAGEMENT NARRATIVE\n\nFull section: `references/template-and-sections.md` — read it before building the document.\n\n## 06 — FURTHER DUE DILIGENCE\n\nFull section: `references/template-and-sections.md` — read it before building the document.\n\n## CITATIONS — superscript markers and a numbered source list\n\nFull section: `references/template-and-sections.md` — read it before building the document.\n\n## NOT IN THIS VERSION\n\nTwo things belong in the next version and are named here so they are not silently dropped: a\nconsensus evolution chart showing how estimates moved around the event, and a market reaction\nsection covering the share price move on the print and on the call. Neither runs by default now.\n\nFull section: `references/template-and-sections.md` — read it before building the document.\n\n## THE SURPRISES IN FULL — the section behind the page-one table\n\nFull section: `references/template-and-sections.md` — read it before building the document.\n\n## THE LENGTH BUDGET — a hard constraint\n\nFull section: `references/writing-standards.md` — read it before writing the first section.\n\n## RETRIEVAL COMPLETENESS — before writing \"not retrievable\"\n\nFull section: `references/writing-standards.md` — read it before writing the first section.\n\n## THE MATERIALITY PASS — after the body is built, before anything is handed back\n\nFull section: `references/writing-standards.md` — read it before writing the first section.\n\n## HARD RULES\n\nFull section: `references/hard-rules.md` — read it before the first retrieval.\n\n## BUILD MAP\n\nWork through the elected sections in order. Write each section to the report as\nit completes; a run stopped part way should leave a usable debrief, not nothing.\n\n1. Ground the event: the call date, participants by role, format, whether the\n   transcript is complete, and the release and guide in force. This is retrieval\n   for your own use and it renders as the four-line provenance strip, not as a\n   section. Detail beyond those four lines goes to the appendix.\n2. Split the transcript into prepared remarks and question and answer before\n   reading either. Record where the split falls.\n3. Pull the release and the guide in force so the delta can be established; label\n   release-sourced findings as such.\n4. Build the elected numbers work: consensus, actuals, guide, and the trailing\n   four-period table — reported plus the prior three comparable periods.\n5. Build the elected segment work: per-segment driver split where disclosed,\n   structural shift, and geography — never inferring volume from revenue and price.\n6. Read the prepared remarks against the trailing three-period baseline; mark\n   inflection points only where a dated change in direction is evidenced.\n7. Read the question and answer: cluster by theme, count questioners per theme,\n   then score answers.\n8. Build \"what to watch next\": dated catalysts mapped to the thesis, named\n   sensitivities, channel and order-book commentary, and exactly five questions.\n9. Assemble \"what the call changed\" last among the sections, since it draws on\n   all of them.\n10. Write the conclusion.\n11. Run the self-check.\n\nIf a retrieval block returns nothing, write the section with the gap named and\nmove on. Never stall on one line.\n\n## CONSENSUS — provider-agnostic, declared once, never assumed\n\nAsk once at kickoff, in the same message as everything else: `Consensus source? Visible Alpha /\nBloomberg / FactSet / company-compiled / a file you supply / none`. Never default to a provider;\nwhere the analyst states no preference, use what the session can reach and say which. Record the\nprovider and its as-of date on every consensus figure. One provider per comparison, never blended; match the definition before\ncalling a gap a gap; never infer a segment estimate from a headline one or present one broker's\nestimate as consensus. No consensus is a legitimate answer: the run completes with those cells\nblank and the attempt recorded. Full rule: `references/samaya-conventions.md` § CONSENSUS.\n\n## RETRIEVAL AND CITATION\n\nRoute every pull by what the item is:\n\n- **The call itself** — the transcript via Samaya, or the uploaded file.\n- **Reported actuals, guidance language, segment figures** — the release,\n  presentation and filings via Samaya.\n- **Consensus** — the elected source, as of a stated date.\n- **Market data** — as of a stated date, and labelled as market data.\n- **Prior calls, for comparison work** — transcripts via Samaya, each dated.\n\nEvery figure and every quotation carries its source and its as-of date. Where a\nconsensus vintage is stale relative to the call, say how stale.\n\n## DESK — established, never assumed\n\nSell-side, buy-side and sales and trading get the same content; only emphasis and order change.\nEstablish the desk from context; ask once, in the kickoff message, only where genuinely unclear;\notherwise build in the neutral order and say so in the header. Never fork the artefact per desk.\nNever render an empty value — omit the row and name it once. Run the defaults, then say what else\nexists: only the blocking kickoff item blocks. Full rule: `references/samaya-conventions.md`\n§ DESK.\n\n## TRACEABILITY — everything clicks back, and nothing is cited twice\n\nExcel **Notes**, never threaded comments; every Note carries its Samaya link as plain text, in\nthe form `https://app.samaya.ai/app/chat/{chatId}?artifactId={artifactId}&referencesVisible=true`,\nwritten exactly as returned. Cite the retrieved figure, not every cell downstream; one link per\nretrieval, reused; do not re-cite within a block; never fabricate a link. Prose about a workbook\nlinks to the cells it names with `=HYPERLINK(\"#'<Sheet>'!<Cell>\", \"<label>\")`, resolved by row\nlabel at runtime. Full rule: `references/samaya-conventions.md` § TRACEABILITY.\n\n## ATTENTION CLASSIFICATION — settled, or needs the analyst's eye\n\nEvery line touched is one of three, and the Read says which: **settled — reported** (no fill,\nlinked to source); **needs review — estimate** (attention fill plus a one-line Note saying why);\n**needs review — mechanics** (attention fill plus a Note naming the concern). The attention fill\nis the only fill, so it needs no legend; if more than a tenth of the touched lines want it, say so\nin the Read instead of colouring the block. Full rule: `references/samaya-conventions.md`\n§ ATTENTION CLASSIFICATION.\n\n## AUTHORSHIP AND HOUSE STYLE\n\nAuthorship is Samaya AI, never the operator: `Samaya AI` on every Note, on the document\nproperties and in footers and mastheads; inherited creator values cleared; a third party's\nauthorship on their own content left intact. Styling tokens live in `samaya-brand`, which wins\nover anything restated here; flag a missing token rather than substituting. Full rule:\n`references/samaya-conventions.md` § AUTHORSHIP AND HOUSE STYLE.\n\n## DOCUMENT ARCHITECTURE — fixed, and identical in every run\n\nMasthead → headline (one declarative sentence naming the finding, written last) → the takeaway →\nkey-numbers tiles → contents (where more than three sections ran) → body in menu order → appendix\n→ conclusion. Nothing that reports an absence appears above the body; the provenance strip is six\nlabelled items at most and the only block above the takeaway; a source-class legend on page one\nand a running footer on every page. Full layout and the page-one gates:\n`references/document-architecture.md`. The strip here is: event,\nparticipants and format; comparison basis; source completeness — four labelled lines, with\ntranscript renderings that disagree, figures omitted for conflict and market-data as-of lines in\nthe appendix.\n\n## THE TAKEAWAY — page one, and it states what is new\n\n**The first sentence names what this call changed, not what this report covers.** A takeaway that\nopens by describing the event, the quarter or the structure of the document has wasted the only\nlines guaranteed to be read. The test is simple: **could this sentence have been written from the\nrelease alone?** If yes, it is not the takeaway — find the thing that needed the call.\n\nWhere the differentiated finding is qualitative, lead with it in the speaker's own words and then\nsay what it implies. A conceded diagnosis, a reframed cause, a position defended more weakly than\nlast quarter: these lead over a beat.\n\n## THE TAKEAWAY — the paragraph itself\n\nNot a paragraph on this skill. Four labelled parts, in this order, the whole block\nfitting the lower half of page one:\n\nFull section: `references/template-and-sections.md` — read it before building the document.\n\n## CONCLUSION — mandatory, five sentences, always present\n\nEvery run closes with a five-sentence conclusion, regardless of scope, sections\nor depth. Prose, no bullets. Each sentence has a job:\n\n1. **What the call changed** — the single most consequential difference between\n   what was known before the call and what is known after it.\n2. **Answered most fully** — the theme answered with the most specificity, and what\n   made it specific.\n3. **Not answered** — the theme deflected, declined or answered thinly, named with\n   the count of times it was raised.\n4. **What the call did to the guide** — whether the guide hardened, softened or\n   held, and where it now sits against consensus.\n5. **What settles the open question** — the disclosure or event that resolves the\n   thing the call left open, with its date where scheduled.\n\nWrite it last, from the sections above. Where the transcript was partial,\nsentence one says so before anything else. It reports a call; it issues no\nrating, target or recommendation.\n\n## OUTPUT\n\nMarkdown report, built to the document architecture above without deviation:\nmasthead, headline, takeaway, contents, body, appendix, conclusion. Tables for\nthe numbers against the print, the trailing four-period frame and the segment\ndecomposition; prose for the question-and-answer read, where the surrounding\ncontext is what makes an answer's quality legible. The appendix carries the\ntranscript provenance and any missing portions, consensus vintage and staleness,\nmarket-data as-of lines, and the not-run list.\n\n## VALIDATION GATES\n\nPlus the common gates in `references/samaya-conventions.md` and the page-one gates in\n`references/document-architecture.md`.\n\nFull section: `references/validation-gates.md` — read it before handing back.\n\n## SCOPE\n\nRuns on a single listed company's results event — the release and the call\ntogether — any sector, any listing, any period. Works from a transcript retrieved\nthrough Samaya or one the user uploads, with the release and prior periods pulled\nthrough Samaya. Reports the reported numbers on a trailing frame and what the call\nchanged against them. It does not value the company,\ndoes not build or edit a model, and does not issue a rating, a target or a\nrecommendation.\n"
}

SHA-256 of public snapshot: caec868ae75a40029c6c45d98fde6ffd1cc2d98754acde980e4574fe06ba1ecc