v0.7.4

What's Changed

• fix(server): reject cross-site and DNS-rebound requests in open mode by @0xMassi in #208
• chore(release): 0.7.4 by @0xMassi in #209

Full Changelog: v0.7.3...v0.7.4

v0.7.3

What's Changed

• fix(fetch): let a body that keeps arriving finish past the total budget by @0xMassi in #205
• fix(pdf): hold one parsed copy of a PDF at a time by @0xMassi in #206
• fix(mcp): load tools when Claude Code falls back from server/discover by @0xMassi in #204
• chore(release): 0.7.3 by @0xMassi in #207

Full Changelog: v0.7.2...v0.7.3

v0.7.2

What's Changed

• Release 0.7.2: render through the hosted API in prebuilt MCP builds by @0xMassi in #202

Full Changelog: v0.7.1...v0.7.2

v0.7.1

What's Changed

• docs(readme): swap in NodeMaven's October banner and link by @0xMassi in #192
• fix(reddit): read Reddit from its public Atom feeds by @0xMassi in #196
• Extractor, fetch and Anthropic client fixes by @0xMassi in #199
• fix(core): keep the segments React streams in hidden blocks by @0xMassi in #197
• Release 0.7.1 by @0xMassi in #200

Full Changelog: v0.7.0...v0.7.1

v0.7.0

What's Changed

• docs: add Thordata as a Studio Partner by @0xMassi in #128
• fix(cli): request the full cloud result when a diff or LLM step follows by @0xMassi in #129
• docs(readme): swap in NodeMaven's September banner and link by @0xMassi in #130
• ci(windows): check the example.com title, not its markdown by @0xMassi in #139
• fix(fetch): one retry path that honors Retry-After by @0xMassi in #133
• fix(proxy): reject SOCKS proxies instead of advertising them by @0xMassi in #134
• feat(core): drop hidden elements from extracted content by @0xMassi in #135
• chore(bench): one-command benchmark against open-source extractors by @0xMassi in #136
• feat(crawl): respect robots.txt by default by @0xMassi in #137
• fix(core): make stripped hidden nodes unselectable by @0xMassi in #140
• feat(crawl): worker-pool crawler that sees links, redirects and Crawl-delay by @0xMassi in #141
• feat(crawl): crash-safe crawls with streamed jsonl and periodic checkpoints by @0xMassi in #142
• feat(render): local JavaScript rendering with --render auto by @0xMassi in #143
• feat: render JavaScript pages in crawls, batches and the MCP server by @0xMassi in #144
• feat: pick values and records from pages with CSS selectors by @0xMassi in #145
• feat: keep only what answers a question (--query) by @0xMassi in #146
• feat(mcp): serve over streamable HTTP with a bearer token by @0xMassi in #147
• feat: screenshots in the CLI and the MCP server by @0xMassi in #148
• feat: every browser profile by name; fix the Safari profile by @0xMassi in #149
• fix(cli): --raw-html exits 1 on HTTP errors by @0xMassi in #151
• docs: fix cargo doc warnings, deny them in CI, refresh CLAUDE.md by @0xMassi in #152
• feat: opt-in disk cache for fetched pages by @0xMassi in #150
• feat: XPath 1.0 selection in the CLI, MCP and core by @0xMassi in #153
• feat: block images, fonts or media while rendering by @0xMassi in #155
• feat: data tables as JSON in the CLI, MCP and core by @0xMassi in #154
• feat: find elements by text and find every element like an example by @0xMassi in #156
• feat: query-guided crawling in the CLI, MCP and core by @0xMassi in #157
• feat: split pages into chunks for retrieval by @0xMassi in #160
• feat: Python package bundling the CLI by @0xMassi in #159
• fix(fetch): keep rendered iframes off private addresses by @0xMassi in #158
• feat: page actions for local rendering by @0xMassi in #161
• feat: save a rendered page as a PDF by @0xMassi in #162
• feat: get past Cloudflare challenges in the local browser by @0xMassi in #165
• fix(fetch): let executeJavascript bodies await before they return by @0xMassi in #166
• feat: capture the network responses of a rendered page by @0xMassi in #164
• feat: keep a persistent browser profile between runs by @0xMassi in #163
• chore(bench): measure selector drift after site redesigns by @0xMassi in #167
• feat(fetch): present every local render as regular Chrome by @0xMassi in #169
• feat(core,cli,mcp): markdown with numbered link citations by @0xMassi in #170
• fix(fetch): follow a crawl redirect that lands on its own page by @0xMassi in #168
• feat: save a page as MHTML or its responses as WARC by @0xMassi in #171
• fix(fetch): space crawl requests per host instead of per worker by @0xMassi in #173
• feat: stop a query crawl once the answer looks found by @0xMassi in #174
• feat: list a site's URLs from the Common Crawl index with --map by @0xMassi in #172
• fix(fetch): close every launched Chrome on SIGTERM by @0xMassi in #175
• fix(mcp): add the crawl fallback to the map tool by @0xMassi in #176
• fix(bench): take a bins dir as the second argument in every script by @0xMassi in #177
• chore(release): 0.7.0 by @0xMassi in #179
• fix: findings from the 0.7.0 release review by @0xMassi in #181
• fix(bench): leave robots compliance unscored when robots.txt cannot be read by @0xMassi in #182
• test(fetch): wait until the wrapper script can run before using it by @0xMassi in #183
• chore: name the licence AGPL-3.0-only by @0xMassi in #184
• docs: record the licence as AGPL-3.0-only in CLAUDE.md by @0xMassi in #185
• docs: name the licence AGPL-3.0-only in the READMEs by @0xMassi in #186
• release: 0.7.0 by @0xMassi in #180

Full Changelog: v0.6.23...v0.7.0

v0.6.23

What's Changed

• fix(release): reconcile v0.6.22 metadata and gates by @0xMassi in #121
• docs: refresh NodeMaven sponsor links and offers by @0xMassi in #125
• docs: replace NodeMaven sponsor banner by @0xMassi in #126
• Prepare v0.6.23 extraction and client changes by @0xMassi in #127

Full Changelog: v0.6.22...v0.6.23

v0.6.22

Highlights

Added

• Added support for an optional hosted model provider while preserving existing provider priority.
• Improved hosted-service model discovery so managed deployments do not probe local inference services they do not use.

Fixed

• MCP requests made before initialization now return the accurate invalid-parameters error when required negotiation data is missing.
• Encoded characters are preserved while parsing documents and sitemaps.
• Valid PDFs whose header appears at the end of the permitted prefix are accepted.

Security

• Updated document parsers and added practical bounds for pathological PDFs.

Contributors

• @XiaoHuo888-hue made their first contribution in #116.

Full changelog: https://github.com/0xMassi/webclaw/blob/v0.6.22/CHANGELOG.md

Full comparison: v0.6.21...v0.6.22

v0.6.21

What's Changed

• docs(pdf): clarify EmptyPdf condition by @ayushcodes13 in #111
• docs: finish the EmptyPdf wording sweep by @0xMassi in #113
• chore: stop CI fetching 1000 live sites, drop a dead workflow, refresh the MCP manifest by @0xMassi in #114
• fix(mcp): upgrade rmcp to 3.1 so an unreadable line no longer ends the session by @0xMassi in #112
• chore(release): 0.6.21 by @0xMassi in #115

New Contributors

• @ayushcodes13 made their first contribution in #111

Full Changelog: v0.6.20...v0.6.21

v0.6.20

What's Changed

• fix(mcp): reply -32601 to pre-handshake requests instead of exiting by @0xMassi in #108
• chore(release): 0.6.20 by @0xMassi in #110

Full Changelog: v0.6.19...v0.6.20

v0.6.19

What's Changed

• perf(sitemap): skip fallback probes when robots.txt declares sitemaps by @0xMassi in #101
• chore(release): 0.6.19 by @0xMassi in #102

Full Changelog: v0.6.18...v0.6.19