v0.5.1

🐛 Bug Fixes

• vercel: drop the duplicate homepage Link route (#34) (c53d8d0)
• vercel: keep the catch-all rewrite off cached routes (#33) (7bdc497)
• vercel: keep the raw prefix out of the section roots (6bf81d2)

Full Changelog: v0.5.0...v0.5.1

v0.5.0

🚀 Features

• app-side discovery registry and a details slot for llms.txt (#29) (facb88f)
• defaults: negotiate markdown for 8 more AI agents (#28) (5665273)

Full Changelog: v0.4.0...v0.5.0

v0.4.0

🚀 Features

• locale roots as homepages, twin Link pair in the Vercel hit phase (#26) (fa3d88a)

🐛 Bug Fixes

• sections: ignore headings inside fenced blocks (#27) (d6e83b1)

Full Changelog: v0.3.0...v0.4.0

v0.3.0

🚀 Features

• append the resources block to a source-backed / document (#22) (2e82fa4)

🐛 Bug Fixes

• prerender: hint each page's raw twin from its own response (#23) (97a311c)
• prerender: skip site-owned, redirecting and missing twins (#24) (230c7d6)

Full Changelog: v0.2.0...v0.3.0

v0.2.0

🐛 Bug Fixes

• follow-up pass from the nuxt.com migration (#21) (ae5d886)
• treat disabled companion modules as absent and type their hooks (#20) (4eecec0)

Full Changelog: v0.1.2...v0.2.0

v0.1.2

🐛 Bug Fixes

• follow-up correctness pass from the stack review (#19) (d62204e)
• llms: make list() optional and gate the full-document fallback (#11) (ecd7835)
• runtime correctness pass (#10) (9095fed)
• vercel: carry the canonical Link pair onto prerendered twins (#15) (92d91b7)
• vercel: collect late route rules and scope edge routes to GET/HEAD (#9) (9e43599)

Full Changelog: v0.1.1...v0.1.2

v0.1.1

🐛 Bug Fixes

• openapi: dedupe operation ids against the caller's (#7) (bc7bf82)
• set Vary on the markdown representations (#6) (bd99433)

Full Changelog: v0.1.0...v0.1.1

v0.1.0

🚀 Features

• let sites extend the discovery layer (c636719)
• llms: generate llms.txt from the content adapter (cf61a77)
• mcp: export the pieces an agent docs tool is built from (9d044cc)
• module: give excludePrefixes the extend / replace shape (c0761f3)
• module: re-export AgentListEntry and AgentSectionSelector (43a1a01)
• openapi: contribute the discovery layer as OpenAPI fragments (e11c713)
• openapi: describe the MCP endpoint alongside its server card (2ca66cc)
• openapi: give every generated operation a stable operationId (841b506)
• raw: let agent-discovery:index set the title and description (497965c)
• sitemap: group sitemap.md into sections (db499fb)
• sitemap: keep the raw prefix out of @nuxtjs/sitemap (8952be9)
• skills: serve agent skills with a generated index (1d42d3e)

🐛 Bug Fixes

• alias #agent-discovery app-side so server routes typecheck (1bac937), references #agent-discovery
• comark: render the same document the content adapter does (1d171df)
• content: stringify with the minimark @nuxt/content resolves (0566da6)
• content: strip the highlighter style node from raw markdown (59a796f)
• discovery: cache the documents that cannot change between builds (bf9593b)
• llms: absolutise same-origin links in llms.txt (74a4a6f)
• llms: keep pages without a markdown body out of the bridge (c200a7c)
• mcp: detect the toolkit at modules:done (1b0db45)
• module: copy the default arrays into the negotiation config (78b207f)
• module: detect the companion modules at modules:done (be178af)
• module: exclude /sitemap.md whenever its link is registered (3b7b448)
• module: only cache-redirect the rules the routes negotiate (0545aa2)
• playground: prepare the module before the vercel build (d93b321)
• pre-publish correctness pass (#5) (e30e0d0)
• raw: omit an empty title or description from the frontmatter (e06fc02)
• robots: contribute through the robots:config hook (68f55c1)
• sitemap: resolve the markdown twins through rawUrl (abd2dd1)
• skills: quote the description holding a colon (095e714)
• skills: rename migrate skill (8947016)
• skills: say why a SKILL.md frontmatter could not be read (f758ba5)
• types: keep the module's own source type-checking inside a consuming site (ca8ccc1)
• vercel: honour Accept: text/markdown;q=0 at the edge (6ddc1aa)

Full Changelog: v0.0.1...v0.1.0