Write Mintlify Changelog

Write a Mintlify changelog

Writes and audits entries in a Mintlify changelog page. Enforces a reader-facing voice modeled on Stripe, Linear, Resend, and Mintlify: short, benefit-led, concrete, no vanity.

This skill owns changelog voice only. When the official mintlify capability is available, use it for current <Update> syntax and other Mintlify mechanics. Otherwise inspect the site's existing components and current Mintlify documentation instead of guessing.

Locating the changelog

The changelog usually lives at docs-mintlify/changelog.mdx. If that path does not exist, look for a changelog.mdx or changelog/*.mdx referenced in the docs.json navigation, and confirm with the user before editing. This skill operates only on the Mintlify changelog page, never on a release-please CHANGELOG.md (see Scope).

The core rule

Every sentence must answer: "what can a reader now do, or rely on?" If a line describes how the team built something, cut it.

Writing rules

1. Length. One sentence for minor changes. Two short sentences or up to 6 bullets for major releases. Anything longer is two entries pretending to be one.
2. Voice. Active, present tense, reader as subject. "You can now filter by date" or "Listening endpoints now include apple_music_url", not "We added date filtering" or "Date filtering has been implemented".
3. Benefit before mechanism. Lead with the capability. Put the tool/endpoint name second. "Full article text via new get_article tool", not "New Voyage AI integration enables get_article".
4. Concrete nouns. Name the endpoint, field, tool, or parameter explicitly (wrap in backticks). Vague ("improvements to reading") is useless.
5. Link out. For anything with its own doc page, link it. Do not re-explain in the changelog.
6. No emojis. No exclamation marks. Project-wide rule and matches Stripe/Linear convention.

Anti-patterns, always cut

• Version archaeology. "Rolls up v0.3.0 to v0.4.3 into one entry", "this release combines...". The date label does this work.
• Vanity metrics. Coverage percentages, "rescues NYT/Reuters", "3x faster", "now supports 12 providers". Readers care about the outcome, not the work. If a number matters (a new cap, a new limit), state it flatly, no celebration.
• Future work. "Coming soon", "planned for next release", "future: Goodreads support". Changelogs are for shipped things only. File futures in a roadmap doc.
• Implementation trivia. Embedding model names, algorithm constants (RRF k=60, cosine, 512 dim), tokenizer details, cache TTLs, internal class names (verify_credentials method), framework specifics. Exception: if a reader can pass or observe the thing, it stays.
• Counting the work. "12 new tools", "10 entity resources", "3 new prompts": numbers without named items are filler. Either name the useful ones or omit.
• Self-reference to prior entries. "As mentioned last release", "building on the March update". Each entry stands alone.
• Hedged language. "We've begun rolling out", "some users may see". Either it shipped or it did not.

Structure conventions

Frontmatter, the <Update> block

<Update label="<Month Day, Year>" tags={["<Tag>", ...]}>

Tags, pick from this closed set:

• New releases: a whole product or service ships (MCP server, docs site, a new domain)
• New features: capabilities added to an existing product (new endpoint, new field)
• Improvements: changes to existing behavior (raised limits, better matching, perf)
• Fixes: bug fixes (rarely used; fold into Improvements unless user-visible)

Combine tags only when the entry genuinely spans them.

Heading

## <Feature or domain name>

• No version numbers in headings, except when the shipping unit is a versioned package readers install (then ## MCP server v0.4.3 is OK, but prefer bare ## MCP server even then, and let the entry's date carry the timeline).
• Domain: feature pattern for scoped updates: ## Reading domain: delta sync, ## Watching: reviews and ratings.
• Title case for the main noun, sentence case for the rest.

Body

Two shapes. Pick one.

Shape A, prose lead plus bullets (for a debut or multi-item release):

## <Feature>

<One-sentence summary of what shipped and who it's for.>

- **<Capability>** — <one sentence, names the endpoint/tool/field>.
- **<Capability>** — <one sentence>.

Shape B, pure bullets (for a tight set of related changes):

## <Feature>

- <Capability> (`endpoint` or `tool_name`).
- <Capability> (`endpoint`).

Shape C, bare prose (for single-item updates, up to 3 sentences):

## <Feature>

<What it does. How to use it, if non-obvious. Link out.>

Separator style

• Use em-dash — between a noun and its gloss: `get_article` — returns the full article body.
• Never -- (double hyphen) in prose. Never - either outside of list markers.
• Markdown list markers are - (single hyphen plus space). Sub-bullets indent two spaces.

Note: this is the one place this changelog style uses the em dash, as a gloss separator inside entries. Body documentation generally uses simpler punctuation.

Bullet shape

• Bold the capability name when using prose plus bullets (Shape A): - **Semantic search** — ....
• Skip bolding in Shape B (pure bullet lists of endpoints/tools).
• End every bullet with a period.

Audit procedure

When asked to audit, run this checklist on every <Update> block in the changelog:

1. Length. Longer than about 8 bullets or 2 paragraphs? Candidate for trimming or splitting.
2. Preamble. Any "rolls up", "combines", "this release" framing? Cut.
3. Future work. Any "coming soon", "planned", "future"? Cut.
4. Vanity. Any percentages, multipliers, coverage stats, or "rescues X / Y"? Cut.
5. Counts. Any "N new tools / resources / prompts" without the items named? Replace with named items or cut.
6. Implementation trivia. Any model names, algorithm constants, internal methods, infra details the reader cannot observe or invoke? Cut.
7. Heading. Version number that is not a shipped package? Strip.
8. Tags. Are tags from the closed set (New releases, New features, Improvements, Fixes)? Normalize.
9. Separators. Any -- in prose? Replace with —.
10. Voice. Any passive voice or "we did X" framing? Rewrite to reader-facing.
11. Links. Any referenced endpoint or feature that has its own doc page but is not linked? Add the link.

Report findings as a per-entry punch list before applying fixes. Apply in one pass.

Examples

Good (matches rules)

<Update label="March 16, 2026" tags={["New features"]}>

## Apple Music enrichment

Listening endpoints now include `apple_music_url` and `preview_url` fields on
tracks, artists, and albums. Links go directly to the matching item on Apple
Music. Preview URLs point to 30-second audio clips when available.

</Update>

Why it works: one capability, names the fields, explains the two URLs in one sentence each, no work-talk.

Bad (rewrite)

## Reading sync improvements

We're excited to announce that we've rolled out a major overhaul of the
Instapaper sync pipeline, now using the `have` and `highlights` API parameters
for delta syncing (~95% bandwidth reduction). Also added a `verify_credentials`
method to the Instapaper client for credential health checks. Coming soon:
Goodreads support.

Violations: preamble ("We're excited"), vanity metric (95%), internal method name (verify_credentials), future work (Goodreads).

Rewrite:

## Reading domain: delta sync

Instapaper sync now transfers only new or changed bookmarks and highlights each
cycle, and deletions propagate from Instapaper. `/health/sync` now includes the
Reading domain.

New entry process

1. Read the release's diff or summary. Pull out reader-facing changes only.
2. Pick the tags from the closed set.
3. Pick a shape (A/B/C) based on how many items there are.
4. Draft the heading using domain:feature if the change is scoped, else a noun phrase.
5. Write each bullet / paragraph against the rules.
6. Run the audit checklist on the draft before inserting.
7. Insert at the top of the changelog (entries are reverse chronological, right under the frontmatter and title).

Scope

This skill operates only on the Mintlify changelog page (typically docs-mintlify/changelog.mdx). Release-please-generated CHANGELOG.md files are out of scope: they serve a different audience (developer release log), are derived from conventional commits, and should not be hand-edited.