stanfish06 Peer Analysis Skill
Generates a professional HTML report (styled like a Moody's peer analysis PDF) for a target
company and up to 3 credit peers. Data is pulled from multiple Moodys MCP server MCP tools and
consolidated into a single HTML artifact covering peers comparison, ratings chart, and ESG.
The workflow is single-artifact file delivery: gather all data, assemble the complete HTML
document in memory, write it to disk as a standalone .html file, and present it to the user for
download. No fenced code blocks, no streaming partial sections, no JSON payload, no progressive
edits. The downloaded file is the sole deliverable.
⚠️ CRITICAL — NON-NEGOTIABLE OUTPUT CONTRACT
The LLM MUST write the final report to disk and present it as a downloadable file. This is the only acceptable form of delivery for this skill. Specifically:
- The LLM MUST write the complete, standalone HTML document (
<!doctype html>→</html>) to/mnt/user-data/outputs/{target_company_slug}_peer_analysis.htmland then callpresent_filesto surface it to the user.- The LLM MUST NOT print the raw HTML source in the chat — no
```htmlfenced code blocks, no partial snippets, no inline HTML in the assistant message.- The LLM MUST NOT split the report across multiple files or multiple messages.
- The LLM MUST NOT substitute prose, Markdown, JSON, or links for the file download.
- If data gathering fails partially, still write and deliver the file with best-available content and
"--"placeholders for missing cells — never skip the file delivery.Treat any other output shape as a hard failure of the skill.
Required MCP servers
Moodys MCP server — tools used: findEntity, getEntityPeers, getEntityRatings,
getEntityCreditOpinion, getEntityFinancials (sections: Profile, Summary, RatingOutlook, FactorsLeadingToUpgrade,
FactorsLeadingToDowngrade, CreditStrengths, CreditChallenges, ESGConsiderations,
KeyIndicatorsTable, ScorecardTable), getEntityEsg, getEntitySectorOutlook
ProbabilityOfDefault - Prod - Moodys MCP - NEW URL — PD tools:
getEntityProbabilityOfDefault— retrieve Current PD, 1-year PD, 5-year cumulative PD, and implied rating for the target company AND each peer. Used to build the PD comparison panel in Section 2. Setpd_comparisontonullin the JSON payload if data is unavailable for the target or fewer than 2 peers.
If any of the tools required for a section do not exist, inform the user: One or more tools required for this section are not available under your current subscription. Unlock more of the expert insights, data, and analytics you trust. Get Link:https://www.moodys.com/web/en/us/capabilities/gen-ai/ai-ready-data.html with us to learn more.
Bundled files
assets/template.html— self-contained static report shell (CSS + named section placeholders, including pre-shaped tables for a fixed target+3-peer layout). No embedded data, no inline script. Treat this file as the read-only structural reference: read it, fill it in mentally, and emit the complete filled document in the final response.
Template (shared)
Before emitting the HTML report, read both:
skills/shared/template/SKILL.md— authoring rules (which classes / snippets are owned by the shared layer, allowed per-skill overrides, outlook-badge usage).skills/shared/template/assets/template.html— canonical CSS (inside<style id="shared-template-css">) and literal HTML markup snippets (inside<template>tags) for the document head, cover, TOC, section block, sources-section wrapper, footer, and outlook-badge.
Lookup order — always check the shared template before inventing. If a class, design token,
layout primitive, or scaffold element you need is not defined in this SKILL.md or already
present in this skill's assets/template.html, the shared template skill is authoritative. Do
not invent CSS, HTML scaffolds, or design tokens that the shared skill already provides; do not
silently restyle anything the shared skill owns (cover, TOC, section, sources-section wrapper,
footer, outlook-badge, design tokens, reset, body / page base).
At emit time, copy the contents (not the <style> wrapper) of <style id="shared-template-css">
from the shared asset into the parent template's reserved marker region between the CSS-comment
markers /* BEGIN shared-template-css ... */ and /* END shared-template-css */. For HTML
scaffolds (head boilerplate, cover, TOC, sources-section wrapper, footer), use the literal markup
from the matching <template> snippet in the shared asset. The parent template no longer
carries duplicated chrome CSS — those rules ship only in the shared asset.
This skill uses the cover-multi variant. Skill-specific override retained above the
marker region: **.page { max-width: 1050px }** (PA reports are wider than the 900px canonical
default to fit the multi-column scorecard / key-indicators tables). Skill-specific CSS that
stays local: .sub-heading, all PA-specific table classes (.pa-table, .credit-drivers-table,
.ki-table, .sc-table), and the chart helpers (.chart-container, .chart-title,
.chart-legend, .chart-legend-item, .chart-legend-swatch). All outlook-badge usage in this
skill must use the canonical pastel variants (stable / positive / negative / review /
na) defined by the shared skill — no solid-fill or inline-color overrides. PA flags the target
entity on the cover with class="company-chip target"; the chip wrappers themselves come from
the shared cover-multi snippet.
Citations (shared)
Before emitting any [n] reference inline, any per-section recap block, or the end-of-document
Citations block, read both:
skills/shared/citations/SKILL.md— authoring rules (numbering, hyperlinking, source data shape, carve-outs).skills/shared/citations/assets/template.html— canonical CSS (inside<style id="shared-citations-css">) and literal HTML markup snippets (inside<template>tags) for inline references, the end-of-document Citations block, and the optional.section-citationsrecap.
At emit time, copy the contents (not the wrapper) of <style id="shared-citations-css">
from the shared asset into the parent template's reserved marker region, located inside
assets/template.html between the CSS-comment markers
/* BEGIN shared-citations-css … */ and /* END shared-citations-css */. The parent
template no longer carries duplicated citation CSS — those rules ship only in the shared
asset.
The prefix used for the end-of-document container in this skill is pa, so the container id
is #pa-sources. Optional per-section recap blocks live in #pa-cite-peers,
#pa-cite-ratings, and #pa-cite-esg. Internal MCP tool names (e.g. getEntityCreditOpinion)
are NEVER rendered inside .source-meta.
Step 1 — Resolve the target company
Call findEntity with the company name provided by the user. Store the canonical entity name
and entity ID.
If the user provides only one company name, that is the target company. The peers will be discovered automatically in Step 2.
Step 2 — Discover and resolve peers
Call getEntityPeers for the target company. Select the top 3 peers returned.
You now have a set of up to 4 companies (1 target + up to 3 peers). For each peer, call
findEntity to resolve its canonical entity name and ID.
Step 3 — Read the template
Read assets/template.html (relative to this skill directory) once. Keep its exact structure —
CSS, <head>, section order, table skeletons (including all pre-shaped 4-company columns/rows),
row labels, and element IDs — as the scaffold for the final artifact. Do not copy it to the
workspace and do not open it.
Step 4 — Gather all data in parallel
For every company in the set (target + up to 3 peers), fire all of the following in a single parallel batch (one message, many tool calls):
Credit Opinion data (per company)
Call getEntityCreditOpinion,getEntityFinancials requesting these sections, about the key Indicators table this should be populated with most recent fiscal year available (e.g. 2025 FY) on getEntityFinancials. Not LTM allowed:
CRITICAL —
getEntityFinancialsis the authoritative source for all quantitative data. Always callgetEntityFinancials(withexcludeInterimData: true) for every entity in the peer set to obtain the most recent fiscal year-end figures. Do NOT rely solely on the Key Indicators or Scorecard exhibits embedded ingetEntityCreditOpinion— those may reference an older period. After fetching financials, identify each entity's latest year with a non-null Revenue value and use that year exclusively for Key Indicators charts, Scorecard "Current" column, and the Revenue bar chart. LTM figures are not permitted.
| Section parameter | Purpose |
|---|---|
Profile |
Company description for the Peers Table |
FactorsLeadingToUpgrade |
Upgrade factors for Credit Drivers |
FactorsLeadingToDowngrade |
Downgrade factors for Credit Drivers |
CreditStrengths |
Credit strengths for Credit Drivers |
CreditChallenges |
Credit challenges for Credit Drivers |
KeyIndicatorsTable |
Financial metrics for Key Indicators table |
ScorecardTable |
Scorecard data |
Ratings (per company)
Call getEntityRatings — retrieve the current long-term rating, rating class, rating date,
outlook, and historical ratings (need at least the last 5 rating actions for the Ratings Chart).
ESG (per company)
Call getEntityEsg — overall CIS classification plus E, S, G sub-scores.
Sector outlook (once per sector)
Call getEntitySectorOutlook for the target company's sector. Reuse for peers in the same
sector.
Probability of Default (target + all peers)
Call getEntityProbabilityOfDefault for the target company AND every resolved peer in a
single parallel batch. Retrieve:
- 1-year PD (%)
- 5-year cumulative PD (%)
- PD-implied rating
Compute the peer average for 1-year PD and 5-year cumulative PD (exclude the target).
Determine pd_signal:
"red"— target 1-year PD is >2× peer average"amber"— target 1-year PD is 1.25–2× peer average"green"— target 1-year PD is ≤1.25× peer average
Build pd_gap_1yr label (e.g. "+0.079pp above peer avg (3.5×)").
Set pd_comparison to null if the PD tool is unavailable or data is missing for the target
or fewer than 2 peers — the panel is suppressed automatically.
Hold all results in context for Step 5 synthesis.
Step 5 — Synthesize + emit the complete artifact
After data is gathered, assemble the complete filled HTML document and write it to disk, then present it to the user for download (see Step 7 for the file-naming and delivery contract).
The assembled document must:
- Be a complete, standalone HTML document (doctype →
</html>) that renders without external dependencies. - Preserve the template's
<head>(CSS, fonts), section order, table skeletons, row labels, and element IDs exactly. Only the empty targets defined below are populated.
Write in professional financial/credit analysis language. Always reference specific companies by name. The target company is always the first column / first row in every table. Render order of content inside the code block follows the page top-to-bottom so the artifact is human-readable as well as browser-renderable: cover/header fields first, then Peers Comparison sub-sections (1–6), Ratings Chart, ESG, and citations/sources last.
Attribute substantive claims with numbered citation references. The exact inline markup, the
URL-less fallback, and the rule that n matches the row position of the source inside
#pa-sources are defined in skills/shared/citations/SKILL.md.
Cover / header fields (write first)
#pa-report-date— e.g.April 20, 2026(plain text)#pa-footer-date— same value (plain text)#pa-target-company— canonical name of the target (plain text)#pa-peer-count— integer count of peers resolved in Step 2 (plain text)#pa-company-chips—<span class="company-chip target">Target</span>followed by one<span class="company-chip">Peer</span>per peer (space-separated; target first)#pa-cover-img-right,#pa-cover-img-bottom— optional. If you have image URLs or data URIs to use, set thesrcattributes and add thehas-cover-imageclass to the corresponding container (<div class="cover-top has-cover-image">and/or<div class="cover-bottom has-cover-image">). If you do not have images, leave the template as-is — the empty image strips will collapse automatically and the cover will render as a clean navy block with the accent bar.
Section 1 — Peers Comparison Table & Analysis
1) Peers Table → <tbody id="pa-peers-table">
Write one <tr> per company. Two columns: company name + a substantive description paragraph
sourced from the Credit Opinion Profile section (headquarters, business overview, key
brands/segments, approximate TTM revenue). Description paragraphs may carry inline citations.
2) Peers Rating → <tbody id="pa-peers-rating">
Write one <tr> per company. Three columns: company name, rating (formatted as
{rating} ({rating class} / {date})), and <span class="outlook-badge …">Outlook</span>.
Use class rules below.
3) Credit Drivers — 5 rows × 4 company columns, pre-shaped <table> with per-cell IDs
Write the 4 company header cells first:
#pa-cd-col-1— target name (plain text)#pa-cd-col-2,#pa-cd-col-3,#pa-cd-col-4— peer names in order (leave empty if fewer than 3 peers were resolved)
Then write the 16 body cells (#pa-cd-r{1..4}-c{1..4}). Each cell is a substantive paragraph
with specific quantitative thresholds where available. Cells may carry inline citations:
- Row 1 (
Upgrade factors) — fromFactorsLeadingToUpgrade - Row 2 (
Downgrade factors) — fromFactorsLeadingToDowngrade - Row 3 (
Credit strengths (qualitative)) — fromCreditStrengths - Row 4 (
Credit challenges (qualitative)) — fromCreditChallenges
Row 5 ("Quantitative support") has been removed. Do NOT emit a 5th row in the Credit Drivers table. The pre-shaped table in
assets/template.htmlhas 5 row slots — leave#pa-cd-r5-c{1..4}blank and suppress the row label for Row 5 by setting its<th>to an empty string or by omitting it entirely if the template allows. Quantitative metrics are covered by the Key Indicators bar charts immediately below.
If a Credit Opinion section is missing for a company, write Not available in the corresponding
cell rather than leaving it blank.
4) Key Indicators — Peer Bar Charts → #pa-ki-charts
Replace the static Key Indicators table with a set of horizontal peer bar charts matching
the visual style from the sector-drift report. Emit a <div id="pa-ki-charts"> block
containing five peer-bar-chart-outer panels, one per metric:
| Chart | Metric | Unit | Source | Note on "lower is better" |
|---|---|---|---|---|
| 1 | FCF Margin % | % |
(OCF − |Capex|) / Revenue | Higher = better |
| 2 | Debt / EBITDA | x |
Moody's pre-computed Key Indicators | Lower = better |
| 3 | EBITDA Margin % | % |
EBITDA / Revenue | Higher = better |
| 4 | EBITA / Interest | x |
Moody's pre-computed Key Indicators | Higher = better |
| 5 | RCF / Net Debt | % |
Moody's pre-computed Key Indicators | Higher = better |
For each chart:
- Show one bar per company (target first, then peers sorted by descending value, then AVERAGE last)
- The target bar uses
background:#0066cc(blue); AVERAGE bar usesbackground:#2e7d32(green) - Peers use
background:#7ca8d8(steel blue) for positive values,background:#e0a0a0for negative - Draw an orange median line (
background: #e87722) at the peer median position - Show the fiscal year for each company's bar by appending the FY label in parentheses to
the
pbc-nameelement (e.g.Target Co (FY2025),Peer A (FY2024)). This is required because different entities may have different most-recent fiscal years. Use the same FY year identified viagetEntityFinancialsfor each entity. - Display the fiscal year reference at the top of the block (e.g.
FY2025, % of revenue | Orange line = peer median | Green bar = peer average); if entities have mixed FYs, writeMost recent FYinstead of a single year - Include a note line at the bottom with: peer median, peer avg (n=X), company value, gap
Compute peer average and peer median across all resolved peers (exclude the target).
FY period header: show the most recent common FY year at the top of #pa-ki-charts in
a <div class="ki-fy-header"> (e.g. KEY INDICATORS — FY2025 | Company vs. Peers).
Reference snippet for one chart panel (emit five of these inside #pa-ki-charts):
<div id="pa-ki-charts">
<div class="ki-fy-header">KEY INDICATORS — FY2025 | Company vs. Peers</div>
<div class="peer-bar-chart-outer">
<div class="pbc-title">FCF MARGIN — COMPANY VS. PEERS</div>
<div class="pbc-subtitle">FY2025, % of revenue | Orange line = peer median | Green bar = peer average</div>
<div class="pbc-rows">
<!-- Target (subject) row -->
<div class="pbc-row">
<div class="pbc-name subject" title="Target Co">Target Co</div>
<div class="pbc-track">
<div class="pbc-median-line" style="left:52.3%"><div class="pbc-median-tip">Median 8.4%</div></div>
<div class="pbc-fill subj-neg" style="left:0;width:12.1%"></div>
</div>
<div class="pbc-val neg">-2.1%</div>
</div>
<!-- Peer rows (sorted descending by value) -->
<div class="pbc-row">
<div class="pbc-name" title="Peer A">Peer A</div>
<div class="pbc-track">
<div class="pbc-median-line" style="left:52.3%"></div>
<div class="pbc-fill peer-pos" style="left:0;width:100%"></div>
</div>
<div class="pbc-val pos">12.3%</div>
</div>
<!-- AVERAGE bar (last, separated by dashed border) -->
<div class="pbc-row" style="margin-top:6px;padding-top:6px;border-top:1px dashed #d0d7e3">
<div class="pbc-name average">AVERAGE</div>
<div class="pbc-track">
<div class="pbc-fill avg-bar" style="left:0;width:70.7%"></div>
</div>
<div class="pbc-val avg-val">8.7%</div>
</div>
</div>
<div class="pbc-note">Peer median 8.42% | Peer avg 8.70% (n=6, Moody's). Company: -2.10%. Gap vs. median: -10.52pp.</div>
</div>
<!-- Repeat for Debt/EBITDA, EBITDA Margin %, EBITA/Interest, RCF/Net Debt -->
</div>
Bar width calculation — for charts with only positive values:
- Find
maxAbs= max of all absolute values across company + peers fillW=(value / maxAbs * 88).toFixed(1) + '%'
For charts that could include negative values (FCF Margin):
- Use a zero-anchored track:
zeroPct = 44, positive bars grow right, negative bars grow left fillW = (|value| / maxAbs * 44).toFixed(1) + '%'fillLfor negative:(zeroPct - |value|/maxAbs*44).toFixed(1) + '%'
Median line position: medPct = (medV / maxAbs * 88).toFixed(1) + '%' (positive-only charts)
Source all values from getEntityFinancials (most recent FY — same period used in scorecards).
If a value is missing, omit the bar row for that company.
5) Scorecards — 7 rows × 8 sub-columns (4 companies × Current/Forward), pre-shaped <table>
Write the 4 top-level company headers first:
#pa-sc-col-1— target name#pa-sc-col-2,#pa-sc-col-3,#pa-sc-col-4— peer names
Write the 8 sub-header labels:
#pa-sc-col-{1..4}-curr— current-period label that must exactly match the most-recent FY period used for that entity in the Key Indicators bar charts (e.g. if Apple's KI data is from FY2025, writeFY2025; if Microsoft's is from FY2024, writeFY2024). Each entity's label is set independently — do not force a single common year across all columns.#pa-sc-col-{1..4}-fwd— forward-period label (e.g.Forward)
FY LABEL SYNCHRONY — STRICTLY ENFORCED (Sections 4 and 5): The fiscal year label shown in Section 4 (Key Indicators bar charts) for each entity MUST be identical to the label shown in the
#pa-sc-col-{n}-currsub-header for that same entity in Section 5 (Scorecards). This is a hard consistency requirement:
- Determine the most-recent non-null full fiscal year-end for each entity from
getEntityFinancials(withexcludeInterimData: true). Label it asFY{YYYY}(e.g.FY2025,FY2024). Never use "LTM", "TTM", "Current", or any period label other than the specificFY{YYYY}form.- Write that exact
FY{YYYY}string in:
- The
pbc-nameelement of each company's row inside every Section 4 bar chart (appended in parentheses, e.g.Apple Inc. (FY2025)), AND in the<div class="ki-fy-header">block label.- The
#pa-sc-col-{n}-currsub-header cell for that entity in Section 5.- The data values themselves — all numbers in Section 4 and Section 5 Current cells must be sourced from the same
FY{YYYY}period.- If two entities have different most-recent FYs, each gets its own correct label independently. The header
<div class="ki-fy-header">should then readKEY INDICATORS — Most Recent FY | Company vs. Peers.- This synchrony check is mandatory before emitting the final HTML. Mismatched labels between Section 4 and Section 5 are a hard failure of the skill.
~(tilde) prefix rule — STRICT: The~prefix (e.g.~2.5x,~15%) means approximately estimated / forecast and must only appear in Forward (-fwd) cells. It must never appear in Current (-curr) cells. Current-period cells contain actual reported or Moody's-adjusted figures sourced fromgetEntityFinancials; use the precise value as-is (e.g.2.3x,14.8%). If the exact current figure is unavailable, writeN/A— do not substitute a tilde-prefixed estimate.
Then write the 56 body cells (#pa-sc-r{1..7}-c{1..4}-curr and -fwd) for the 7 rows:
- Row 1 — Scale: Revenue/Sales (USD bn)
- Row 2 — Profitability: EBIT(A) margin
- Row 3 — Leverage: Debt/EBITDA
- Row 4 — Cash flow: RCF/Net debt
- Row 5 — Coverage: EBIT(A)/Interest
- Row 6 — Scorecard-indicated outcome
- Row 7 — Actual rating
Source from ScorecardTable in the Credit Opinion and getEntityFinancials
6) Conclusion → #pa-conclusion
Three <p> paragraphs focused on differentiation of the target among its peers. Paragraphs
may carry inline citations:
- Paragraph 1: How the target differentiates positively vs. weaker peers on quantitative metrics (leverage, coverage, cash flow). Cite specific ratios.
- Paragraph 2: Business risk profile comparison — where the target is more concentrated or less diversified vs. stable peers. Reference upgrade/downgrade triggers.
- Paragraph 3: Practical positioning — frame the target's "path to stand out" by linking back to the upgrade/downgrade framework and key risk factors.
Section 2 — Ratings Chart (± Probability of Default)
SECTION 2 TITLE RULE — STRICTLY ENFORCED: The heading for Section 2 (in both the TOC
<li>and the<div class="section-heading">) must be set conditionally at emit time:
- If PD data is available (i.e.
pd_comparisonis NOTnull) → use "Section 2. Ratings Chart & Probability of Default"- If PD data is unavailable (i.e.
pd_comparisonisnull, the MCP server is unreachable, or data is missing for the target or fewer than 2 peers) → use "Section 2. Ratings Chart"The template ships with the longer title as a placeholder; overwrite it in both locations (
#sec2 .section-headingand the matching<li>in<ul class="toc-list">) whenever PD data is absent. Never show "& Probability of Default" in the title if there is no PD panel.
Chart → #pa-ratings-chart
Emit a single inline <svg> block authored by the agent (see SVG template below). The chart
plots the last 5 rating actions for each company as a line chart. X-axis uses sequential labels
(Rating 1 … Rating 5, most recent first); Y-axis maps Moody's rating symbols to a numeric
scale.
PD Panel → #pa-pd-panel
Immediately after the ratings chart, emit the Probability of Default panel as a horizontal
bar chart using the same .peer-bar-chart-outer style as the Key Indicators charts in
Section 1. If PD data is unavailable, leave #pa-pd-panel empty (the element collapses).
Emit one peer-bar-chart-outer block showing the 1-year PD (%) for each company. Use the
same CSS classes and bar-width formula as the KI charts (pbc-fill, pbc-track, pbc-rows,
etc.). Include a note row at the bottom with peer median, peer average, and the gap signal.
Signal colouring for the target bar:
pbc-fill subj-neg(red) — target 1-year PD > 2× peer average (pd_signal: "red")pbc-fill subj-pos(blue) — all other cases
Reference snippet (emit this block directly into #pa-pd-panel):
<div class="peer-bar-chart-outer">
<div class="pbc-title">PROBABILITY OF DEFAULT — 1-YEAR PD (%)</div>
<div class="pbc-subtitle">Moody's PD model | Orange line = peer median | Green bar = peer average</div>
<div class="pbc-rows">
<!-- Target (subject) row — blue or red depending on pd_signal -->
<div class="pbc-row">
<div class="pbc-name subject" title="Target Co">Target Co</div>
<div class="pbc-track">
<div class="pbc-median-line" style="left:38.5%"><div class="pbc-median-tip">Median 0.032%</div></div>
<div class="pbc-fill subj-neg" style="left:0;width:100%"></div>
</div>
<div class="pbc-val neg">0.111%</div>
</div>
<!-- Peer rows (sorted descending by pd_1yr) -->
<div class="pbc-row">
<div class="pbc-name" title="Peer A">Peer A</div>
<div class="pbc-track">
<div class="pbc-median-line" style="left:38.5%"></div>
<div class="pbc-fill peer-pos" style="left:0;width:54.1%"></div>
</div>
<div class="pbc-val pos">0.060%</div>
</div>
<!-- AVERAGE bar (last, separated by dashed border) -->
<div class="pbc-row" style="margin-top:6px;padding-top:6px;border-top:1px dashed #d0d7e3">
<div class="pbc-name average">AVERAGE</div>
<div class="pbc-track">
<div class="pbc-fill avg-bar" style="left:0;width:28.8%"></div>
</div>
<div class="pbc-val avg-val">0.032%</div>
</div>
</div>
<div class="pbc-note">Peer median 0.032% | Peer avg 0.032% (n=3). Target: 0.111%. Gap: +0.079pp above peer avg (3.5×). PD-implied rating: A1.</div>
<div class="pbc-note" style="margin-top:4px;font-style:italic">One-sentence PD narrative goes here.</div>
</div>
Bar width calculation: fillW = (pd_1yr / maxPD * 88).toFixed(1) + '%' where maxPD is
the maximum 1-year PD across all companies including the target.
Median line position: medPct = (medianPD / maxPD * 88).toFixed(1) + '%'
Analysis → #pa-ratings-analysis
One <p> paragraph on rating trajectories — which companies show improvement, deterioration,
or stability, and what that implies for relative credit positioning. May carry inline citations.
Section 3 — ESG Table & Analysis
ESG Table → <tbody id="pa-esg-table">
Write one <tr> per company. Five columns: company name, Overall (CIS-), Environmental (E-),
Social (S-), Governance (G-).
Analysis → #pa-esg-analysis
Two <p> paragraphs. Paragraphs may carry inline citations:
- Paragraph 1: Compare ESG profiles across the peer set. Identify which companies share similar profiles and what drives the overall classification.
- Paragraph 2: Highlight the key differentiator (typically Governance) and explain which company stands out as weaker/stronger and why.
Citations & sources (write last)
#pa-cite-peers— optional<div class="section-citations">…</div>recap for the peers-comparison section. Use the markup defined in skills/shared/citations/SKILL.md. Omit if empty.#pa-cite-ratings— optional recap for the ratings-chart section. Same markup.#pa-cite-esg— optional recap for the ESG section. Same markup.#pa-sources— end-of-document Citations rows. One<div class="source-item">per source, in[1], [2], …order, using the canonical row markup from the shared citations skill. Internal MCP tool names are NEVER rendered in.source-meta.
Streaming protocol (element → content mapping)
| Element ID | Content type |
|---|---|
#pa-report-date |
Plain text date |
#pa-footer-date |
Plain text date (same value) |
#pa-target-company |
Plain text (canonical target name) |
#pa-peer-count |
Plain text integer |
#pa-company-chips |
<span class="company-chip target"> + <span class="company-chip">s |
#pa-cover-img-right, #pa-cover-img-bottom |
<img> element — set src attribute |
<tbody id="pa-peers-table"> |
<tr> rows (company name + description) with inline citations |
<tbody id="pa-peers-rating"> |
<tr> rows with <span class="outlook-badge …"> |
#pa-cd-col-1 … #pa-cd-col-4 |
Plain text (company names) |
#pa-cd-r{1..4}-c{1..4} |
Prose cell content (short <p> or plain text) with inline citations |
#pa-ki-charts |
Five .peer-bar-chart-outer panels (FCF Margin, Debt/EBITDA, EBITDA Margin, EBITA/Interest, RCF/Net Debt) |
#pa-sc-col-1 … #pa-sc-col-4 |
Plain text (company names, span 2 sub-columns) |
#pa-sc-col-{1..4}-curr, #pa-sc-col-{1..4}-fwd |
Plain text sub-header labels (Current / Forward period) |
#pa-sc-r{1..7}-c{1..4}-curr, …-fwd |
Plain text scorecard cell values |
#pa-conclusion |
Three <p> paragraphs with inline citations |
#pa-ratings-chart |
Single <div class="chart-container">…<svg>…</svg>…</div> block |
#pa-pd-panel |
PD comparison panel — horizontal bar chart using peer-bar-chart-outer style |
#pa-ratings-analysis |
One <p> paragraph with inline citations |
<tbody id="pa-esg-table"> |
<tr> rows (company + overall + E + S + G) |
#pa-esg-analysis |
Two <p> paragraphs with inline citations |
#pa-cite-peers, #pa-cite-ratings, #pa-cite-esg |
<div class="section-citations">…</div> (optional, see shared citations skill) |
#pa-sources |
<div class="source-item"> rows (see shared citations skill) |
If fewer than 3 peers are returned, leave unused column/row IDs empty. Do not collapse or remove cells — the pre-shaped table simply keeps those cells blank.
Reference HTML snippets
Peers table row (<tbody id="pa-peers-table">):
<tr>
<td class="company-name">Target Co</td>
<td>Headquartered in Dearborn, Michigan, Target Co is a global automaker with TTM revenue of approximately $180B <a href="https://example.com/credit-opinion-target" target="_blank" class="cite-ref">[1]</a>, anchored by its F-Series, SUV, and Pro commercial businesses <a href="https://example.com/profile-target" target="_blank" class="cite-ref">[2]</a>.</td>
</tr>
Peers rating row (<tbody id="pa-peers-rating">):
<tr>
<td class="company-name">Target Co</td>
<td>Baa3 (Senior Unsecured - Dom Curr / 2024-05-09)</td>
<td><span class="outlook-badge stable">Stable</span></td>
</tr>
Company chips (#pa-company-chips):
<span class="company-chip target">Target Co</span>
<span class="company-chip">Peer One</span>
<span class="company-chip">Peer Two</span>
<span class="company-chip">Peer Three</span>
ESG table row (<tbody id="pa-esg-table">):
<tr>
<td class="company-name">Target Co</td>
<td>CIS-3</td>
<td>E-4</td>
<td>S-3</td>
<td>G-2</td>
</tr>
Credit Drivers — per-cell write (repeat for each of the 20 body cells):
Leverage sustained below 2.5x Debt/EBITDA, EBITA margin above 7%, and sustained positive free cash flow after dividends <a href="https://example.com/credit-opinion-target" target="_blank" class="cite-ref">[1]</a>.
Conclusion paragraph (inside #pa-conclusion, same inline-citation pattern applies to
#pa-ratings-analysis and #pa-esg-analysis):
<p>On pure quantitative credit metrics, Target Co stands out decisively: Debt/EBITDA of 0.7x and EBITA/Interest of 27x are in line with the strongest-scoring peers, while its scale advantage (LTM revenue of $294B vs. $61B and $23B for the peer set) and FCF/Debt of 41% add a differentiation layer the others cannot match <a href="https://example.com/credit-opinion-target" target="_blank" class="cite-ref">[1]</a><a href="https://example.com/peer-ratings" target="_blank" class="cite-ref">[2]</a>.</p>
Key Indicators — single row (emit the full <tr> for each company inside the
#pa-key-indicators table, keeping the per-cell id attributes intact):
<tr>
<td id="pa-ki-r1-company">Target Co</td>
<td id="pa-ki-r1-v1">LTM (30 Sep 2025)</td>
<td id="pa-ki-r1-v2">180.2B</td>
<td id="pa-ki-r1-v3">155.6B</td>
<td id="pa-ki-r1-v4">7.8B</td>
<td id="pa-ki-r1-v5">4.33%</td>
<td id="pa-ki-r1-v6">13.1B</td>
<td id="pa-ki-r1-v7">11.88x</td>
<td id="pa-ki-r1-v8">3.62x</td>
<td id="pa-ki-r1-v9">136.0B</td>
<td id="pa-ki-r1-v10">18.5%</td>
</tr>
Scorecard — per-cell write (Current or Forward):
2.7x
Section citations recap (#pa-cite-peers, #pa-cite-ratings, #pa-cite-esg) and
sources rows (#pa-sources): see
skills/shared/citations/SKILL.md for the canonical markup.
Reuse the same [n] numbering across inline references, optional recap blocks, and the
end-of-document Citations block. Never include MCP tool names in .source-meta.
Ratings Chart SVG template
Emit the entire chart as a single HTML string into #pa-ratings-chart. The agent authors the
SVG directly — there is no runtime script.
Hardcoded constants
- Dimensions:
W = 700,H = 350 - Padding:
PAD_L = 70,PAD_R = 30,PAD_T = 30,PAD_B = 50 - Plot area: width
= W - PAD_L - PAD_R = 600, height= H - PAD_T - PAD_B = 270 - Number of x-axis points:
maxPts= the largest number of ratings any company has, capped at 5 - Palette (first = target):
#0066cc,#e6550d,#1a7a4a,#8b5cf6,#d63384
Moody's rating → numeric value map
| Rating | Value | Rating | Value |
|---|---|---|---|
| Aaa | 21 | Ba1 | 11 |
| Aa1 | 20 | Ba2 | 10 |
| Aa2 | 19 | Ba3 | 9 |
| Aa3 | 18 | B1 | 8 |
| A1 | 17 | B2 | 7 |
| A2 | 16 | B3 | 6 |
| A3 | 15 | Caa1 | 5 |
| Baa1 | 14 | Caa2 | 4 |
| Baa2 | 13 | Caa3 | 3 |
| Baa3 | 12 | Ca | 2 |
| C | 1 |
RATING_LABELS (index 0..20 → symbol): ["C", "Ca", "Caa3", "Caa2", "Caa1", "B3", "B2", "B1", "Ba3", "Ba2", "Ba1", "Baa3", "Baa2", "Baa1", "A3", "A2", "A1", "Aa3", "Aa2", "Aa1", "Aaa"].
Coordinate formulae
Compute once per chart:
- Collect every rating value across all companies' last-5 series →
values[]. minV = clamp(min(values) - 1, 1, 21)maxV = clamp(max(values) + 1, 1, 21)(ensuremaxV > minV; if equal, setmaxV = minV + 1)
Per point (i, v) where i is the x-index (0-based) and v is the rating value:
xPos(i) = PAD_L + (i / (maxPts - 1)) * 600=70 + (i / (maxPts - 1)) * 600yPos(v) = PAD_T + 270 - ((v - minV) / (maxV - minV)) * 270=30 + 270 - ((v - minV) / (maxV - minV)) * 270
Reference SVG block
<div class="chart-container">
<div class="chart-title">Last 5 Ratings for Target and Peers</div>
<svg viewBox="0 0 700 350" width="100%" preserveAspectRatio="xMidYMid meet">
<!-- Horizontal gridlines + y-axis rating labels (one line + label per rating tick from minV to maxV) -->
<line x1="70" y1="30" x2="670" y2="30" stroke="#e5e7eb" stroke-width="1"/>
<text x="64" y="34" text-anchor="end" font-size="10" fill="#475569">Baa1</text>
<line x1="70" y1="84" x2="670" y2="84" stroke="#e5e7eb" stroke-width="1"/>
<text x="64" y="88" text-anchor="end" font-size="10" fill="#475569">Baa2</text>
<line x1="70" y1="138" x2="670" y2="138" stroke="#e5e7eb" stroke-width="1"/>
<text x="64" y="142" text-anchor="end" font-size="10" fill="#475569">Baa3</text>
<line x1="70" y1="192" x2="670" y2="192" stroke="#e5e7eb" stroke-width="1"/>
<text x="64" y="196" text-anchor="end" font-size="10" fill="#475569">Ba1</text>
<line x1="70" y1="246" x2="670" y2="246" stroke="#e5e7eb" stroke-width="1"/>
<text x="64" y="250" text-anchor="end" font-size="10" fill="#475569">Ba2</text>
<line x1="70" y1="300" x2="670" y2="300" stroke="#e5e7eb" stroke-width="1"/>
<text x="64" y="304" text-anchor="end" font-size="10" fill="#475569">Ba3</text>
<!-- X-axis baseline + tick labels -->
<line x1="70" y1="300" x2="670" y2="300" stroke="#334155" stroke-width="1.5"/>
<text x="70" y="320" text-anchor="middle" font-size="10" fill="#475569">Rating 1</text>
<text x="220" y="320" text-anchor="middle" font-size="10" fill="#475569">Rating 2</text>
<text x="370" y="320" text-anchor="middle" font-size="10" fill="#475569">Rating 3</text>
<text x="520" y="320" text-anchor="middle" font-size="10" fill="#475569">Rating 4</text>
<text x="670" y="320" text-anchor="middle" font-size="10" fill="#475569">Rating 5</text>
<!-- Target company line (palette[0] = #0066cc) -->
<path d="M70,138 L220,138 L370,192 L520,192 L670,192"
fill="none" stroke="#0066cc" stroke-width="2.5"/>
<circle cx="70" cy="138" r="4" fill="#0066cc"/>
<circle cx="220" cy="138" r="4" fill="#0066cc"/>
<circle cx="370" cy="192" r="4" fill="#0066cc"/>
<circle cx="520" cy="192" r="4" fill="#0066cc"/>
<circle cx="670" cy="192" r="4" fill="#0066cc"/>
<!-- Repeat <path>+<circle>s for peers 2, 3, 4 using palette[1..3] -->
</svg>
<div class="chart-legend">
<div class="chart-legend-item"><span class="chart-legend-swatch" style="background:#0066cc"></span>Target Co</div>
<div class="chart-legend-item"><span class="chart-legend-swatch" style="background:#e6550d"></span>Peer One</div>
<div class="chart-legend-item"><span class="chart-legend-swatch" style="background:#1a7a4a"></span>Peer Two</div>
<div class="chart-legend-item"><span class="chart-legend-swatch" style="background:#8b5cf6"></span>Pee
…(truncated)