- the Usage fence becomes two src= references to ExampleRender and
ExamplePurify, filled by summer docs:sync
- the Phase 11.1 SC4 acceptance scan now allows top-level src= fences
in module READMEs (nested ones are still refused) and expects the
api/<name> pages to caption them like docs/ pages
- checkSnippets compares top-level src= fences of module READMEs like
docs/ pages; the README src= refusal is gone
- Sync also rewrites drifted src= fences in modules/<name>/README.md,
still all-or-nothing across pages and READMEs
- fenceAnnotator captions top-level src= fences on every page
- the snippet-readme-src planted violation now plants a drifted README
fence instead of the old refusal
- utilities.md, README.md and CLAUDE.md describe src= in module READMEs
- surf.AssembleRouter returns the router and the compiled handler from
one build, so an mcp.enabled app is not built twice; Assemble keeps
its signature and is implemented on top of it
- serve uses AssembleRouter with a deferred Router.Close
- pier.Pier.Close stops the session reaper and the Redis warm-up; the
router that mounted the endpoint owns it
- ExampleAssembleRouter, routing/testing docs and the surf/pier READMEs
- MemoryStore.Close stops the sweep goroutine; Attempt keeps working
- Router.Close runs what the router owns (the fallback MemoryStore of
BuildRouter), never a store published on the app
- BuildRouter closes the router on every error return
- serve closes its router and installCooler's fallback store on exit;
route:list closes its router
- README and rate-limiting docs describe the ownership
- Register keeps a pristine shallow copy of each pointer-to-struct plugin;
every Activate hands its application a fresh copy, so a second
application no longer rebinds the first one's plugins
- value plugins and pointers to non-struct types are used as is
- party.RegisterFactory for plugins whose init builds maps or pointers
- activate() and its tests are unchanged
- README and plugin/lifecycle docs describe the per-application rule
- TestCrossPluginLayoutOverride: a template override naming the lender's
layout, a customized lender layout, Editor Validate and Preview
- TestCrossPluginLayoutSync (Docker): SyncIndex points the borrower rows'
layout_id at the lender's layout row and keeps it on a second run
- resolveLayoutLocked: after the plugin's own alias and the built-in
aliases, a full layout name already in the catalog resolves to itself;
an unregistered full name fails with a require-the-owner hint
- another plugin's short alias and the short-alias error are unchanged
- TestCrossPluginLayout covers register, send, errors and precedence
- postcard README and docs/services/mail.md describe the rule
- refuse cases for all eight probe payloads (noscript href and handler,
svg style, title and textarea, noembed, xmp, conditional comment)
- send fallback and editor refusal for the three that survive rendering
- escaped_send subtests check svg title/textarea sends with an x/net/html
parser oracle, scripting on and off
- accept cases for conditional comments, quoted CSS, a safe noscript
link and a literal comment opener in an attribute
- unsafeMarkupAt runs the existing HTML reading, then a tag reading in
which attribute quoting alone decides where a tag ends
(NextIsNotRawText after raw-text start tags, comment openers blanked)
- both readings share unsafeAttrs; existing refusals keep their message
and snippet
- svg/style end tag inside a quoted attribute is refused at check, save
and send; README and docs/services/mail.md describe the second reading
scriptURLText took data: followed by any non-space character as a data
URL, so names such as Data:Lab, Big Data:Summit and data:Science were
refused on the Branding form and dropped from mail, contrary to the docs.
A data URL now needs its optional MIME type and the ; or , every data URL
has. javascript: and vbscript: keep matching with whitespace after the
colon, since such a URL still runs.
validateHTML ran (?i)\son[a-z]+\s*= over the rendered document, text
included, so a variable such as "buy one = get one free" stopped its own
message, and it is what made Online=Shop in the footer a send failure.
The tokenizer walk refuses every on* attribute, and since CR-02 it also
reads raw-text element and comment content as markup, so the regex is
dropped. Text shaped like a handler is now sent; such site names are shown
and saved, and the Branding refusal copy no longer mentions onclick=.
Single-URL attributes removed ASCII whitespace and control characters
before looking for a scheme, the list-valued, meta content, style and
comment checks did not, so values="java	script:..." and a tab-split
meta refresh passed. The stripping is factored out as stripURLSpace and
used by every scheme check.
The tokenizer-only walk read noscript, title, textarea, style, xmp and the
other raw-text elements as one text token and never looked inside, so an
unencoded javascript: link in <noscript> or <svg><title> passed the check
and was sent. A mail client renders with scripting off and parses svg
children as markup, so the raw content of those elements, and comment
data (Outlook conditional comments), is now checked again as markup;
nesting deeper than eight levels is refused.
- CR-01 rows for the send path, the decoded scheme check and the admin side;
WR-01 and WR-05 rows with their go test commands
- CR-01 status replaces the open item on the Branding site name
- Manual-Only row for the Branding refusal on sm-demo-app; Wave 0 lists the
three new test files
- the workspace-modfile case copies the planted consumer's go.mod before the
modfile run and compares it after with cmp -s (IN-03)
- drops the git status check on the dirty-sibling fixture, which asserted
against the wrong repository; the replace-line grep stays
- TestMailSiteNameSafety: a javascript: site name is a 422 on site_name in en
and pl and stores nothing; Big Data: Summit and Metadata: Weekly save; the
Branding preview drops an unsafe draft name; a name stored before the check
never refuses a template or layout save or stops a send; a mail.from display
name with a colon shows on the default and system layouts
- the mail screens env reads mail.from from its configuration (default
unchanged), so a mutate hook can set the sender
- check-phase35.sh DOCKER_TESTS requires TestMailSiteNameSafety
- TestURLSchemeCheck: prose and scheme words in titles or queries pass; every
URL-bearing attribute, CSS, meta refresh and each entity-encoded or
whitespace-split script URL is refused with a snippet at the markup; a layout
override with javascript: falls back to the file version with inlining on
and off; the editor refuses it on content_html and CSS URLs on content_css
- TestInlineUnsafeOutput: a failing inlined result sends the style-block
version with one warning per layout
- TestFooterSiteNameLiteral: the footer prints the site name escaped and
literally on the default and system layouts; body stays Markdown
- check-phase35.sh GO_TESTS requires the three tests
- TestSiteNameSend: mail.from "Metadata: Weekly" and Branding "Big Data: Summit"
show in header, footer and text on the default and system layouts, on the
file-version, override and override-failure fallback paths
- TestUnsafeSiteNameDropped: unsafe names are left out on every path with one
value-free warning per source; editor saves and previews are not refused
- TestUnsafeText pins the safe and unsafe texts
- check-phase35.sh GO_TESTS requires them and TestMailSafetySmoke
- the footer partial gains a text argument, printed as a paragraph
through html/template and verbatim in the text part; body stays
Markdown for callers that pass Markdown on purpose
- the default and system layouts pass the copyright line as text, so
*Acme*, Acme <Studio> and [x](url) appear as typed, as in the header
(WR-01)
- README partial table, layouts and Branding paragraphs; docs partials
table and built-in layouts paragraph
- TestMailSafetySmoke/literal_footer
- inlineHTML treats a failing post-inline check like an inliner error:
the un-inlined HTML, which already passed validateHTML, is sent with
its style block and the rate-limited inline:<layout> warning
- mail is never stopped by the inliner, as the docs promise (WR-05)
- README and docs/services/mail.md describe the check before inlining
- TestMailSafetySmoke/inline_unsafe_output
- postcard.UnsafeText: a text a layout prints is unsafe when, escaped,
it fails the rendered-HTML checks or holds a script or data URL at a
scheme boundary ("Big Data: Summit" is safe)
- siteNameFrom/layoutSiteName: an unsafe Branding name falls back to the
mail.from display name, an unsafe one of those gives no name; one
rate-limited warning per source (site_name:branding, site_name:mail.from),
never the value; file version, override, fallback and editor alike
- the Branding preview of an unsafe draft name renders without it
- MailBrandSetting.BeforeValidate: 422 on site_name with
backend::lang.mail.brand.site_name_unsafe (en/pl)
- settings Put localizes a BeforeValidate refusal's phrase keys, as a
controller save does
- TestMailSafetySmoke/unsafe_site_name, TestMailBrandSiteName/unsafe_name_refused