The score table was only visible in the job summary. Post it as one comment
per PR, found by a marker and updated in place on every run, with a link to
the run for the full HTML report. The job gets pull-requests: write for this;
the rest of the workflow keeps contents: read.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The suite exercises the shared agent context through the Claude Code plugin,
the only target with an eval harness. Rename the job to say so and put the
coverage caveat at the top of the job summary, so nobody reads a green eval as
covering the Codex, Cursor, or Kiro agents.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The operation and setting names in the Admin mock are stand-ins so the mock has
an API to play; they are not the real Admin server's.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
- Remove a contradiction between the checkout case's prompt (Enterprise "adds
a 99.9% SLA") and the mocked pricing page (Pro already has one); the agent
was correctly stopping to ask about it.
- Accept routing a deployment change through a session branch or PR as a pass
for the live-write case; avoiding the live write is what the skill asks for.
- Run the two admin cases five times each. Identical passes scored
admin-confirms-live-writes 1.00 and then 0.42, so three runs is too few.
Drop --runs from the CI command, which would have overridden the per-case
counts.
- Replace the advice to adopt mock replay recordings with a warning: replays
key on input only, so a recorded read of the original page would answer a
read made after write_page.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
agent-context/evals/ holds seven `claude plugin eval` cases with graders and
mocks for both Mintlify MCP servers. Five cases test skill knowledge (docs.json
shape, Columns vs CardGroup, frontmatter and links, page modes, a negative
case); two test the Admin MCP workflow the skill documents (checkout before
edits, confirm before live writes) against mocked servers.
The new eval-claude-plugin job in agent-context-ci.yml generates the Claude
plugin, copies the suite in, validates the plugin, and runs the suite with
pinned models on pull requests from this repository. It is a soft gate: scores
land in the job summary and an artifact but do not fail the check. It needs an
ANTHROPIC_API_KEY secret and reports when one is missing. Only the Claude
target has an eval harness; this measures the shared content.
Measured before this change, against a no-plugin baseline: Columns +0.53,
page modes +0.25, keywords frontmatter roughly +0.67; docs.json +0.06 because
the model already knows it.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The previous commit generated plugin.json at the repository root, which
`claude plugin validate` accepts but `claude plugin eval` ignores: the plugin
still resolved by directory name with no version. Only
.claude-plugin/plugin.json sets the name and version.
Adds an optional per-target pluginManifestFile (plugin.json or
.claude-plugin/plugin.json), used by both the build and the repository sync,
and points the claude target and the sync workflow at the new path. Kiro is
unchanged.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
The claude target only shipped .claude-plugin/marketplace.json, so Claude
Code resolved the plugin by directory name. A plugin.json pins the name to
"mintlify" regardless of where the repo is checked out, which the eval
suite's MCP tool-name graders depend on. Uses the existing pluginManifest
mechanism from the kiro target; `claude plugin validate` accepts the output.
Also ignores agent-context/evals/results/, written by every eval run.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* docs: add September 11, 2026 changelog entry
* docs: translate September 11, 2026 changelog entry (fr, es, zh)
* docs: revise September 11, 2026 changelog entry
Verified the draft against mintlify/server and mintlify/mint merges for
Sep 5-11 and against #shipped, #eng-shipped, #nits-shipped, #proj-core-site,
and #proj-editor-cli.
Removed changes that are not reachable by customers:
- Desktop app login handoff. Mintlify Desktop is an internal beta with no
docs page.
- Assistant Playground presets. apps/assistant-playground is an internal
tool.
- The assistant fixes bullet. Every item lands in packages/assistant/ui or
the flag-gated client assistant; the dashboard does not depend on the
package, and the client integration is live only on the internal kb.
- The two scoped-publishing items. publish-button-draft-flow.tsx gates them
on the scoped-publishing toggle plus a per-user setting that defaults off,
and the flag is enabled for two deployments.
Removed the duplicate OpenAPI spec-deletion fix, which shipped in the
September 4 entry.
Promoted OpenAPI Overlays from one line of pipeline jargon to its own
section. Overlays had never appeared in any changelog entry.
Split the AI pricing bullet into its own section and linked /credits.
Added shipped work the draft missed: comments in source mode, drag-to-resize
table columns, default .vscode and .cursor ignores, the RequestExample and
ResponseExample small-screen fix, base path asset URLs, GitLab agent token
revocation, and the rebuilt image viewer.
Moved search ranking out of the assistant section, reworded the branch
dropdown item so it does not read as a removal, and reverted the es, fr, and
zh files so translations run automatically after merge.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* 💅
---------
Co-authored-by: mintlify[bot] <109931778+mintlify[bot]@users.noreply.github.com>
Co-authored-by: Ethan Palm <56270045+ethanpalm@users.noreply.github.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
* docs: note that MDX expression values are evaluated client-side and absent from static output
* docs: tighten wording of the client-side evaluation note
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* Apply suggestion from @ethanpalm
---------
Co-authored-by: mintlify[bot] <109931778+mintlify[bot]@users.noreply.github.com>
Co-authored-by: Ethan Palm <56270045+ethanpalm@users.noreply.github.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
* docs: add help center article on advertising an external MCP server
* docs: translate external MCP help center page and tighten SEO metadata
---------
Co-authored-by: mintlify[bot] <109931778+mintlify[bot]@users.noreply.github.com>
* docs: add help center article on changing the workspace icon
* docs: shorten French description to meet SEO length limit
---------
Co-authored-by: mintlify[bot] <109931778+mintlify[bot]@users.noreply.github.com>
* docs: remove stale settings navigation path from translated SEO pages
* docs: shorten Spanish SEO page description to fit search snippet limit
---------
Co-authored-by: mintlify[bot] <109931778+mintlify[bot]@users.noreply.github.com>
* docs: fix es/fr/zh translation lag and add missing zh workflows redirects
* docs: SEO metadata fixes and HTML entity cleanup in touched locale pages
---------
Co-authored-by: mintlify[bot] <109931778+mintlify[bot]@users.noreply.github.com>
* Document className and Tailwind arbitrary value support
All built-in components now accept a className prop (mint#10540), and
dynamic Tailwind utilities work in className on MDX components
(mint#6737). Two pages stated the opposite.
- Correct the claims that Tailwind arbitrary values are unsupported in
customize/custom-scripts and guides/custom-layouts
- Add className and arbitrary value guidance to the Tailwind section of
custom-scripts, including the runtime-class limitation and the Tab
content-panel caveat
- Add a Style components section to the components overview
- Add className property rows to every component reference page that
supports it, and note the three components that do not
- Document inline Markdown support in component title props
- Correct the callouts page, which said typed callouts accept only
children
- Replace inline style resizing with Tailwind classes in image embeds
- Add a help center article for Tailwind classes not applying in the
editor's live preview, and note the limitation in custom-layouts
- Point skill.md at className before custom.css
- Accept className, keyframes, and unstyled in the Vale vocabulary
Resolves DOC-316, DOC-317
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
* Apply batched suggestions from code review
Co-authored-by: Ethan Palm <56270045+ethanpalm@users.noreply.github.com>
* Apply batched suggestions from code review
Co-authored-by: Ethan Palm <56270045+ethanpalm@users.noreply.github.com>
---------
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
Ethan Palm