` blocks, unique to case studies (and rarely used elsewhere) — get them exactly right, the build pipeline and stylesheet depend on them:\n\n**`attribution-quote`** — a pull quote inside the body, attributed to a named speaker:\n\n```\n
\n\n*Italicized verbatim quote, exactly as the customer said it.*\n\n**Speaker Name**\n\nTitle, Company\n
\n```\n\n- The quote line is italic (`*...*`), the name is bold (`**...**`), the title/company line is plain text. Blank lines between each are required for the markdown to render correctly.\n- Title and company can be one line (`Sr Manager Systems Engineering, Fastly`) or, when the company is already obvious from context, just the title.\n- Use 2-4 of these per case study, spaced through the narrative at the moments they land hardest — right after describing the pain, right after the \"why Fleet\" reasoning, and in the results section. Don't front-load them all in one place.\n\n**`checklist`** — a plain list of short items with no markdown bullet syntax, rendered as a checklist by the stylesheet:\n\n```\n
\n\nFirst item — a criterion, a feature, or a quantified result\n\nSecond item\n\nThird item\n
\n```\n\n- Items are separated by blank lines, not `-` or `*` bullets — the div's CSS supplies the checkmark styling.\n- Use it for evaluation criteria (Faire's three MDM-selection priorities) or a short, punchy results list (Foursquare's cost/time savings). Longer results lists (4+ items with more explanation) can also work as plain markdown bullets in the results section prose, as Thumbtack's and Primo's \"With Fleet, [Company] has:\" lists do — either is fine, pick whichever reads cleaner for the story.\n\n### Voice and honesty guardrails — stricter than other content types\n\nCase studies carry more risk than an article: every claim is attributed to a real person at a real company, and it gets fact-checked by that company before or after publishing.\n\n- **Quotes are verbatim, never paraphrased or invented.** If you don't have the customer's exact words for a moment the story needs, write the narrative around it without a quote, or flag it for the author to source a real quote. Do not manufacture a plausible-sounding quote and attribute it to a named person.\n- **Numbers come from the customer, not from you.** Migration percentages, device counts, cost savings, timeframes — these all must trace back to what the customer or the Fleet team working with them actually said. Flag any number you're inferring rather than quoting/sourcing directly.\n- **Name Fleet's own team members' contributions honestly** (e.g., a CSM's role in a migration) only if that's part of the customer's account, not invented color.\n- Otherwise, the standard `content-style` rules apply in full: sentence case headings, no em dashes, no hyperbole (\"revolutionary,\" \"seamless,\" \"game-changing\"), Oxford commas, active voice. Case study prose reads slightly more narrative/conversational than a guide, closer to an article, but it's still restrained — let the customer's own quotes carry the enthusiasm; the narrator's voice around them stays measured and factual.\n\n### Endmatter — case-study-specific meta tags (build-enforced)\n\nBeyond the standard article endmatter (`articleTitle`, `description`, `publishedOn`, `authorFullName`, `authorGitHubUsername`, `category`), the website's markdown build (`website/scripts/build-static-content.js`) requires additional tags for `category: case study` pages, and will fail the build if they're missing:\n\n- **`summaryChallenge`, `summarySolution`, `summaryKeyResults`** — required. These populate the sidebar \"Challenge / Solution / Key results\" summary box on the case study page (`website/views/pages/articles/case-study.ejs`). `summaryKeyResults` **must** be a semicolon-separated list (`\"Result one; Result two; Result three\"`) — a single sentence with no semicolon fails the build.\n- **`companyLogoFilename`** — if present, must reference a file that actually exists in `website/assets/images/`. The build checks this and errors if the file is missing.\n- **`quoteContent`** (the hero pull-quote at the top of the page) — if present, requires `quoteAuthorName`, `quoteAuthorJobTitle`, and `quoteAuthorImageFilename` alongside it, and the image file must exist in `website/assets/images/`.\n- **`companyName`, `companyInfo`** (and optional `companyInfoLineTwo`) — power the \"About [Company]\" sidebar/mobile block. Not build-enforced, but expected by the template; include them.\n\n**Image assets are a hard dependency, not something this skill can produce.** Before the build will succeed, someone needs to add the actual logo and headshot files to `website/assets/images/`, following the site's naming convention (`{descriptor}-{css-width}x{css-height}@2x.{ext}`, e.g. `fastly-logo-104x40@2x.png`, `dan-jackson-120x120@2x.png`). If those files don't exist yet, say so explicitly rather than writing `
` tags that point at nothing — that's a silent build break waiting to happen.\n\nAn escape hatch exists (`
`) for a case study that should render with the plain article template instead of the sidebar layout; it then requires `cardTitleForCustomersPage` instead of the summary/quote/company tags. This is rare — only reach for it if the author specifically doesn't want the sidebar template.\n\n### Related, optional step: testimonials.yml\n\nMany (not all) published case studies also get their pull-quote added as an entry in `handbook/company/testimonials.yml`, which feeds the `
` carousel used across the site (e.g., the `/customers` page). This is a nice-to-have, not build-enforced, and outside this skill's core job — mention it to the author as a follow-up rather than doing it unprompted, since it touches a shared file curated by the marketing team.\n\n## Workflow: writing a new case study\n\n1. Confirm you have real material to draw from — customer quotes (verbatim, attributed), and concrete numbers (device counts, migration timelines, cost figures). If working from raw interview notes or a call transcript, extract the quotes and facts before drafting; don't smooth a real quote into different words.\n2. Identify the three-act shape for this specific story: what was broken, why Fleet, what changed. Pick section headings that fit the story (see canonical examples) rather than reusing one example's headings verbatim.\n3. Draft the body: problem → decision → results, threading 2-4 `attribution-quote` divs at the moments they land, and a `checklist` div for evaluation criteria and/or headline results if the story has a natural short list.\n4. Write the title (sentence case, names the company and the outcome).\n5. Write the full endmatter block: standard article tags, then the case-study-specific summary/quote/company tags. Derive `summaryChallenge`/`summarySolution` as tight paraphrases of the body's first two acts; derive `summaryKeyResults` from the results section, semicolon-separated.\n6. Confirm image assets (`companyLogoFilename`, `quoteAuthorImageFilename`) exist in `website/assets/images/`, or flag that they still need to be added.\n7. Run the `content-style` skill over the prose for voice, sentence case, em dashes, filler, and Fleet terminology.\n8. Run the self-check below.\n\n## Workflow: auditing or editing an existing case study\n\n1. Confirm it's tagged `category: case study`. If it's an `announcements` piece narrating a customer story (like the Deputy example), don't force this structure onto it — flag the mismatch instead.\n2. Read the whole piece and check the three-act shape is intact: problem, decision, results, each grounded in specifics rather than generic praise.\n3. Check every quote reads as something a person would actually say (specific, occasionally imperfect phrasing) rather than smoothed marketing copy — that's often a sign a quote was paraphrased rather than transcribed. Flag for the author to verify against the source if uncertain.\n4. Check the `attribution-quote` and `checklist` divs are syntactically correct (blank lines between quote/name/title, no stray markdown bullets inside `checklist`).\n5. Verify the endmatter has all case-study-required tags and that `summaryKeyResults` is semicolon-separated.\n6. Verify referenced image files (`companyLogoFilename`, `quoteAuthorImageFilename`) actually exist in `website/assets/images/`.\n7. Run the `content-style` skill over the prose.\n8. Run the self-check below, then summarize changes and flag anything you couldn't verify (a number, a quote, a missing image asset).\n\n## Self-check before finishing\n\n- Title is sentence case, names the company, and states the outcome rather than a feature.\n- Body follows the problem → choosing Fleet → results arc, with section headings suited to this story.\n- 2-4 `attribution-quote` divs, correctly formatted (italic quote, blank line, bold name, blank line, title/company), spaced through the narrative rather than clustered.\n- Every quote is verbatim and attributed to a real, named person; no invented quotes.\n- Every number (migration %, timeframe, cost, device count) traces back to the customer or Fleet team, not an inference.\n- A `checklist` div (or a plain bullet list) surfaces the headline results.\n- Endmatter includes `summaryChallenge`, `summarySolution`, `summaryKeyResults` (semicolon-separated), and the company/quote meta tags — or, if using `useBasicArticleTemplate`, `cardTitleForCustomersPage` instead.\n- `companyLogoFilename` and `quoteAuthorImageFilename` reference image files that exist in `website/assets/images/`, or the author has been told they still need to add them.\n- No \"osquery\" in prose (say \"Fleet's agent\" or \"fleetd\"); headings sentence case; no em dashes, filler, or hyperbole.\n- The `content-style` skill has been run over the prose.\n\n## Reference\n\nA blank, copyable skeleton lives at `assets/case-study-template.md`. A structural breakdown of each canonical example lives at `references/canonical-examples.md`.\n\n\n# References Skill\n\nPath: `.claude/skills/fleet-gitops/references/profiles-android.md`\n\n[View Raw Content on GitHub](https://raw.githubusercontent.com/fleetdm/fleet/HEAD/.claude/skills/fleet-gitops/references/profiles-android.md)\n\n\n# References Skill\n\nPath: `.claude/skills/fleet-gitops/references/profiles-apple.md`\n\n[View Raw Content on GitHub](https://raw.githubusercontent.com/fleetdm/fleet/HEAD/.claude/skills/fleet-gitops/references/profiles-apple.md)\n\n\n# References Skill\n\nPath: `.claude/skills/fleet-gitops/references/profiles-windows.md`\n\n[View Raw Content on GitHub](https://raw.githubusercontent.com/fleetdm/fleet/HEAD/.claude/skills/fleet-gitops/references/profiles-windows.md)\n\n\n# References Skill\n\nPath: `.claude/skills/fleet-gitops/references/queries.md`\n\n[View Raw Content on GitHub](https://raw.githubusercontent.com/fleetdm/fleet/HEAD/.claude/skills/fleet-gitops/references/queries.md)\n\n\n# References Skill\n\nPath: `.claude/skills/fleet-gitops/references/software.md`\n\n[View Raw Content on GitHub](https://raw.githubusercontent.com/fleetdm/fleet/HEAD/.claude/skills/fleet-gitops/references/software.md)\n\n\n"}