Quick Start
User: "Compare these two issuer filings and explain the changed risk and MD&A language."
Identify both periods and accessions; compare retrieved passages, gate added/removed claims on symmetric complete coverage, and label metric arithmetic without deriving performance.
Runtime capability checks
Read references/house-rules.md before drafting. "WorkIQ" below names a search capability, not an assumed callable API. Use SearchM365 for M365 content discovery when available and ExecuteSql for supported structured record reads (for example, dated calendar records); inspect the actual tool schema before calling it. Check configured connector capabilities rather than assuming a named provider exists. Public retrieval may use an available direct-fetch tool or the browser; discovery alone is not source evidence. If a capability or destination skill is unavailable, disclose that specific limitation and use the existing upload/gap path, never invent a call or claim it ran. Create files only through the available format skill; otherwise provide the drafted content and state that file creation remains incomplete.
Workflow handoffs — no scope expansion
Only perform the workflow defined in this skill's Scope. For a different job, check that the destination exists and accepts it before handing off; never claim an unavailable skill ran. Do not reinvoke yourself for the current job.
- For a new/ad-hoc institutional questionnaire, use rfp-ddq-response-draft instead.
- For a recurring consultant profile with a prior submission to compare, use consultant-database-refresh instead.
- For named fund/portfolio/strategy periodic investor commentary, use investment-commentary-draft instead.
- For a matched two-filing issuer-language diff, use security-research-brief instead.
- For a fund/share-class product comparison, use peer-product-intelligence instead.
- For a named institutional investment-client mandate review, use institutional-client-review-prep instead.
Scope
Business-workflow boundary: the deliverable is a two-filing evidence diff with matched prior/current passages and retrieval coverage. A numerical variance table or narrative alone is not this workflow. Metrics explain the filing-language changes; they are not a standalone corporate operating-performance commentary.
Do: compare two filings, surface changed language, supply supporting metric context, quote management. Do not: rate, recommend, value, set a price target, forecast, or report partial retrieval as complete.
When NOT to Use
- Standalone corporate financial-variance narrative, including prior-year or prior-quarter comparisons — Finance
variance-commentaryif available. Followreferences/house-rules.md; lack of Finance never turns this into a numbers-only commentary skill. - Competitor fund or product matrix — use
peer-product-intelligenceinstead. - Institutional client review packs — use
institutional-client-review-prepinstead. - Questionnaires or recurring profile refreshes — use
rfp-ddq-response-draftorconsultant-database-refreshinstead. - Single quarter results recap — use
earnings-reviewinstead, where the Capital Markets plugin is installed. - Formatting or restyling an existing document — use the built-in
docxinstead.
Issuer Filing & Earnings Change-Pack
Read an issuer's latest filing against its prior one and report what changed. The value is in the diff, not in restating the filing.
Do: compare filing language, evidence matched passages, supply supporting metric context, quote management. Do not: rate, recommend, value, set a price target, or forecast. This is an input to an analyst's judgment, not a substitute for it.
If the Capital Markets plugin is also installed: that plugin's
company-profile-packbuilds a full standing profile of a company, andearnings-reviewcovers a single quarter's results for a banker audience. This skill is period-over-period change detection for an investment analyst — what moved and what the language stopped saying.Confirm the destination skill exists in this session before handing off. Those skills ship in a different plugin and may not be installed. If the user wants a complete company overview rather than a diff, or a recap of a single quarter's print rather than a filing-language diff, and the destination skill is unavailable: say so, name the plugin that provides it, and offer the closest in-scope alternative — a two-filing diff, or a narrower brief scoped to what this skill covers. Never describe another skill's output as though it ran.
Workflow
1. Identify the issuer and the two filings
Read references/edgar-access.md before opening EDGAR — it covers the retrieval routes,
the ticker lookup, R-pages, the retry rule, and four gotchas that will corrupt figures.
Find the company by ticker, then identify two filings: the latest and the directly comparable prior one (10-K vs. prior 10-K, Q3 vs. prior-year Q3 — not Q3 vs. Q2 unless asked).
State both filings, with period end dates and accession numbers, before reporting any difference. If the prior comparable filing cannot be found, say so — a diff against the wrong period is worse than no diff.
2. Metric context for the filing diff
From the R-pages: revenue, margins, operating income, net income, EPS, segment results, and cash flow. One R-page carries three fiscal years, so the comparison usually costs two or three page loads total.
Include only metrics that support the risk-factor, MD&A, or guidance comparisons below. Report each as prior → current, with the change, with period end dates and a link to the relevant compared passage. Do not produce a separate full financial-variance narrative. If either filing is unavailable, label the diff incomplete rather than substituting a one-period earnings recap or generic operating commentary.
3. Risk-factor diff — the highest-value section
Compare Item 1A across the two filings:
- Added risk factors — usually the most informative signal in the entire filing
- Removed risk factors — language present in the prior filing that is not present in the current one. Report only that it is not present. Do not say the company no longer considers it material: risks disappear through consolidation, restructuring, or incorporation by reference at least as often as through genuine de-risking
- Materially reworded risks — especially escalated or de-escalated language
Quote the changed language. Do not paraphrase a risk factor into your own characterization of severity.
Risk factors sit deep in a 10-K (~1.5 MB). Navigate to Item 1A directly. If you could only retrieve part of it, say so rather than reporting a short list as complete.
Completeness gates apply symmetrically — this is the section's most important rule.
| Claim | Requires |
|---|---|
| A risk was removed | Complete retrieval of both Item 1A sections |
| A risk was added / is new | Complete retrieval of both Item 1A sections |
| A risk was reworded | Both passages actually retrieved and matched |
A truncated read of the current filing looks exactly like a deletion. A truncated read of the prior filing looks exactly like an addition. Both errors are equally wrong and equally easy.
When coverage is partial on either side:
- Do not write
new,added,removed, ordropped. - For something present now, write "currently disclosed; novelty unverified" and say which section was incompletely read.
- Report wording comparisons only for passages you retrieved from both periods, and state explicitly that the comparison is not an exhaustive diff.
Carry coverage status next to every comparison conclusion, not only in the footer — a reader scanning the risk table must see the limitation where the claim is.
4. MD&A and guidance language
What changed in how management explains results. Watch for:
- Guidance raised, lowered, narrowed, withdrawn, or newly introduced
- Language shifting between confident and hedged on the same topic
- A previously discussed initiative that is not mentioned in the current filing — subject to the same completeness rule as removed risks: only report this when both MD&A sections were retrieved in full
- New qualifiers around demand, pricing, or costs
Quote both versions side by side. A change in language is evidence; what it means is the analyst's call. Report the shift; do not interpret intent.
5. Earnings call
Where a transcript is publicly posted on the IR site, summarize prepared remarks and notable Q&A with attribution; use only brief, necessary quotations where permitted, not wholesale reproduction.
Not all issuers post transcripts publicly. If none is available, say
NOT FOUND — no public transcript and use the 8-K earnings release instead. Never reconstruct
what was said on a call from memory or from a secondary summary without marking it
UNVERIFIED — secondary source.
6. Output
Template resolution: user-named file → M365 Cowork Templates/Asset Management/security-research-brief/ → references/brief-outline.md.
Deliverable: .docx or HTML brief, optionally with an .xlsx metric-change table. If the firm
keeps prior research notes in SharePoint, retrieve the last note on this issuer and note
continuity — what the firm flagged last time that did or did not change.
Output length never waives attribution or the output footer: one-line answers, single table cells, inline replies, and short excerpts still carry citations and the footer. If the requested artifact is a spreadsheet, document, or deck, supply content, structure, evidence, markers, and placeholders, then hand creation or formatting to xlsx, docx, or pptx as appropriate.
Source resolution
- Configured MCP
Bloomberg/FactSet/Refinitiv→ consensus estimates, so changes can be read against expectations rather than only against the prior period. Highest-value upgrade for this skill.- Internal research library → prior firm thesis for continuity.
- User upload — the filings or a transcript, when the browser is unavailable.
- Public primary — EDGAR filings, IR site, publicly posted transcripts, FRED.
- Flag —
NOT FOUND/NOT RETRIEVED/NEEDS FEED.
If a configured MCP supplies the item in front of you, use it and do not open the browser for that item. Browsing is for items no connector covers — narrative text, as-reported figures where the accounting basis matters, and anything whose citation needs a primary identifier. A connector being present does not mean it holds the item: if it returns nothing, that is a connector miss, so continue down this list. Never treat it as a dead end, and never report it as the item not existing.
A partial or stale connector answer is a miss for whatever it does not cover. Where the item
is a count, an "all of X" set, an absence claim, or a point-in-time status, a non-empty connector
answer does not end the search — confirm completeness and currency against the authoritative
source, or mark that coverage partial and name what was not covered. A vendor's subset reported
as the whole manufactures a finding that was never established, and a connector never settles a
determination or a point-in-time status on its own.
A connector figure cannot carry a filing citation. This skill cites form type, period end, and accession number. Those come from the filing, so a figure taken from a connector is cited to the connector with its as-of date, never presented as a filing figure.
Consensus estimates are not in EDGAR at any price. Mark them NEEDS FEED — never estimate
what the market expected.
Capabilities. Reading supplied files (PDF, DOCX, XLSX, PPTX, CSV, images with OCR); web retrieval of public sources; Microsoft 365 search where the user has access; configured MCP connectors where present. Artifact creation is delegated — spreadsheets to xlsx, documents to docx, decks to pptx.
When a capability is unavailable: say which one, name what it would have supplied, and continue with what is available. Never silently skip a step and never present an unavailable source as returning nothing. If no source exists for a field, use NEEDS FEED — <source that would supply it>; if a source exists but retrieval failed or access was denied, use NOT RETRIEVED — <source> <failure>; if the source answered and has no record, use NOT FOUND — <source> has no record for <what>.
Guardrails
- A language diff is mandatory. A customer template or connector cannot replace it with standalone corporate variance commentary.
- Output length never waives attribution or the output footer. A one-line answer, single table cell, inline reply, or short excerpt still carries its citation and this skill's footer.
- Never fabricate a figure, quote, date, source, approval status, or document state. Use the skill's gap marker instead.
- If an input page or source document is unreadable, report
NOT LEGIBLE — <document>, page <n>, ask for a cleaner copy, and never guess. - Distinguish unavailable sources exactly: no source exists is
NEEDS FEED; a source exists but retrieval failed or was denied isNOT RETRIEVED; a source answered and has no record isNOT FOUND. Never collapse these outcomes. - Never delete or overwrite existing content, prior submissions, decision sections, reviewer notes, or source files; append, redline, or preserve unless the user explicitly supplies a new blank artifact.
- Format handoffs: spreadsheets go to
xlsx, documents todocx, decks topptx. This skill supplies content, structure, evidence, markers, and placeholders; the built-in creates or formats the file. - If a save, export, or write fails, never claim it succeeded. Report what was produced, where it was and was not written, and what remains incomplete.
- Nothing is sent, submitted, filed, posted, committed to a consultant database, client portal, other portal/database, or emailed by this skill. A draft is a draft and is reported as a draft; the human submits.
- Reuse firm-approved library material only within the firm's authorized rights and this task. For third-party filings, policies, or other protected text, use only brief necessary excerpts where permitted; otherwise summarize with attribution or request authorized source material. Attribution does not grant reproduction rights. Never present third-party wording as the firm's own or relabel a source's classification into another scheme.
- No rating, recommendation, price target, or valuation. Not even hedged.
- No forecast of results, and no characterization of a stock as cheap or expensive.
- Compare like periods; state both filings explicitly.
- Quote changed language verbatim; never paraphrase into a severity judgment.
- Distinguish
NOT FOUND(not disclosed) fromNOT RETRIEVED(page did not load). Retry once first. - Show period end dates, not fiscal-year labels.
- Partial retrieval of a long section is disclosed, never reported as a complete list.
See references/house-rules.md for citation format, gap markers, and source tiers.
Output footer
Output length never waives this footer. A one-line answer, single table cell, inline reply, or short excerpt still carries its citations and the footer below.
Workflow: matched two-filing language diff — metrics are supporting context
Issuer: <name> (<ticker>)
Compared: <form, period end, accession> vs. <form, period end, accession>
Sections retrieved: <full | partial — say which>
Transcript: <source, or "NOT FOUND — no public transcript">
Template: <file used>
Connectors: <MCPs used, or "none — EDGAR reported actuals only">
Gaps: <NOT FOUND / NOT RETRIEVED / NEEDS FEED items>
Internal research aid. Not investment advice, not a recommendation, and not a rating.
Reference files
references/house-rules.md— shared decision gate, attribution and sourcing rules, blocked-vs-absent states, markers, and output footers.references/brief-outline.md— default research brief: issuer framing, reported-actuals tables, filing citations, and gap reporting.references/edgar-access.md— capability-aware SEC EDGAR access, company and accession resolution, and filing-section retrieval. Read before citing a filing.