formatPostTime() used local timezone methods, causing test failure
on UTC CI servers. XHS API timestamps are Beijing time (UTC+8),
so use explicit UTC offset with getUTC*() methods.
- wikiFetch return Promise<unknown> instead of Promise<any>
- Add WikiSearchResult type, remove r: any
- Type wikiFetch responses with inline type assertions
- Only append ... to abstract when actually truncated
Add arXiv (search, paper) and Wikipedia (search, summary) public API adapters.
- arxiv/search: search papers by keyword
- arxiv/paper: get paper details by ID
- wikipedia/search: search articles with lang support
- wikipedia/summary: get article summary
Type safety fixes applied: wikiFetch returns unknown, typed search results.
Co-authored-by: BruceLoveDecimal <39156883+BruceLoveDecimal@users.noreply.github.com>
- Remove fetchCreatorNotesByCdp() and captureNoteDetailApiPayload() raw
WebSocket code (~240 lines) — adapters should use IPage, not raw CDP
- Replace direct CDP WebSocket with IPage.evaluate() in-page fetch
- Fix page: any → IPage in all function signatures
- Simplify to two-tier fallback: API+interceptor → DOM parse
- Rebase onto latest main (resolves cdp.ts/daemon.ts conflicts)
* docs: deduplicate documentation — single source of truth in docs/
- Remove root CDP.md, CDP.zh-CN.md, CLI-ELECTRON.md (now in docs/advanced/)
- Slim adapter READMEs to one-liner + link to docs/ (11 files)
- Update README.md adapter table links to point to docs/
* docs: set VitePress base path for GitHub Pages deployment
URLSearchParams.toString() encodes spaces as +, but Bilibili's WBI
signature verification expects %20. This mismatch causes search
queries with spaces (e.g. "亚马逊 滞销产品") to fail with
TypeError: Failed to fetch due to CORS-blocked error responses.
Fixes#125
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The build manifest includes antigravity/serve which collides with the
hardcoded antigravity serve in cli.ts. Add a guard to skip registry
entries whose subcommand already exists in the site group.
* feat(hf): add top command for hf papers (daily, weekly, monthly)
* feat(footer): add footerExtra support and derive dates from API response
Add footerExtra callback to CliCommand for custom table footer content.
For weekly/monthly periods, derive date range from API response publishedAt
field with local clock fallback.
* fix: truncate long paper titles
* refactor(hf): remove comments column for consistent output
* feat(hf): add --all flag to return all papers
* feat(hf): add paper id column to output
* fix: restore main.ts as bootstrap, sync footerExtra + CDPBridge + domain pre-nav to cli.ts
- main.ts should remain a lightweight entry point delegating to cli.ts
- Preserve CDPBridge fallback (OPENCLI_CDP_ENDPOINT) — PR had hardcoded BrowserBridge only
- Add domain pre-navigation for cookie/header strategies to cli.ts
- footerExtra feature from PR is properly integrated
---------
Co-authored-by: jackwener <jakevingoo@gmail.com>
The old adapter called `search.smzdm.com/ajax/?c=<channel>&s=<q>` which
now returns 404. This caused opencli smzdm search to always return empty
results regardless of keyword.
Fix: navigate directly to `search.smzdm.com/?c=home&s=<keyword>&v=b`
and scrape the rendered DOM via querySelectorAll('li.feed-row-wide').
Also switched from async IIFE to sync IIFE since all data is already in
the DOM after page load — no fetch needed.
Tested: opencli smzdm search --keyword A7M5 returns correct results
with prices and mall names.
Adds 'opencli boss resume --uid <uid>' command that scrapes the chat page
right panel to display candidate resume information including:
- Basic info: name, gender, age, experience, degree, active status
- Work history: time period + company + position
- Education: time period + school + major + degree
- Job being discussed and candidate expectations
Uses UI scraping approach since BOSS Zhipin does not expose a public API
for candidate resume data on the recruiter side.
Add comprehensive Jike (即刻) adapter covering read and write operations.
Read commands:
- user: user posts via m.okjike.com SSR JSON
- topic: topic/circle posts via m.okjike.com SSR JSON
- post: post detail with comments via m.okjike.com SSR JSON
- feed: home timeline via React fiber tree extraction
- search: search posts via React fiber tree extraction
- notifications: notification list via DOM innerText parsing
Write commands (Strategy.UI, browser DOM automation):
- create: publish post via inline compose box
- comment: comment on post via contenteditable paste
- like: like post via _likeButton_ div click
- repost: repost via action bar → popover menu → confirm
Implementation details:
- Three data extraction strategies: SSR JSON, React fiber, DOM manipulation
- Shared JikePost interface and getPostData helper in shared.ts
- All evaluate blocks include try/catch error handling
- Two rounds of parallel Claude + Codex code review applied
Replace ad-hoc string escaping with JSON.stringify() for values
interpolated into JavaScript code strings passed to page.evaluate().
- explore.ts: clickLabels were escaped with only single-quote
replacement, which breaks on labels containing backslashes or
newlines. JSON.stringify() handles all edge cases correctly.
- synthesize.ts: buildEvaluateScript() embedded URLs directly inside
single quotes. JSON.stringify() safely handles URLs containing
special characters.
- boss/chatlist: List chat conversations (招聘端聊天列表)
Uses getBossFriendListV2 API with pagination and job filter support.
- boss/chatmsg: Read chat message history with a candidate
Resolves encryptUid to numeric uid/securityId, fetches via historyMsg API.
- boss/send: Send chat message to a candidate via UI automation
BOSS chat uses MQTT protocol (not HTTP), so this command automates the
web chat UI: clicks on user in list → types in contenteditable editor →
clicks the send button.
All three commands use Strategy.COOKIE and require an active BOSS直聘
login session in Chrome.
The previous approach (nativeSetter + Enter keydown on the search input)
does not reliably trigger Twitter's form submission - the synthetic
KeyboardEvent is ignored by React, leaving the page on /explore with
zero API calls captured.
Use history.pushState + PopStateEvent instead, which triggers React
Router's listener and performs a true SPA navigation to /search.
The interceptor survives because no full page reload occurs.
Tested: "opencli", "it's a test" (single quote), "hello" all return
results with correct author attribution.
Add CLI commands to view Chaoxing assignments and exams by reusing
Chrome login session via the Browser Bridge.
Chaoxing has no flat API for listing assignments/exams. The adapter
follows the browser flow: establish session → fetch course list via
backclazzdata API → enter each course via stucoursemiddle redirect →
click tab to capture iframe URL → navigate and parse DOM.
Commands:
opencli chaoxing assignments [--course <name>] [--status] [--limit]
opencli chaoxing exams [--course <name>] [--status] [--limit]
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix(intercept): use evaluate() for IIFE wrapping in installInterceptor/getInterceptedRequests
Root cause: daemon migration changed these methods from this.evaluate()
to direct sendCommand('exec'), losing the wrapForEval() IIFE wrapping.
CDP received bare arrow functions that were never invoked.
Fixes#98
* fix(twitter): SPA navigation, data path, and author resolution for INTERCEPT commands
- followers/following: install interceptor on profile page, then click
followers/following link (SPA navigation preserves JS context).
Use JSON.stringify for targetUser to prevent injection. Throw on
navigation failure. Update selector: /verified_followers.
- notifications: install interceptor on home, then pushState+popstate
to /notifications. Validate navigation URL.
- search: fix author resolution (core.screen_name, not legacy).
- All: fix GraphQL data path (remove extra .data level), update author
resolution to try core.screen_name before legacy.screen_name.
- followers: remove erroneous .filter(r => r?.url) — interceptor stores
response body JSON, URL filtering happens at capture time.
Replace eager close-window (which caused race conditions when
parallel commands shared the window) with an idle-based timer:
- Window auto-closes 30s after the last command completes
- Each incoming command resets the idle timer
- Consecutive commands reuse the same window (faster)
- No race conditions with parallel execution
- Close-window action kept for explicit cleanup if needed
- Add 'close-window' action to extension protocol and background.ts
- Add Page.closeWindow() method to send close-window command
- browserSession() now closes automation window in cleanup
- Remove domain pre-navigation + 2s wait from main.ts (CDP handles
cross-domain cookies natively, no same-origin workaround needed)
- Net effect: commands run faster, no stale windows left behind
- Delete unused extension/src/executor.ts (chrome.scripting experiment)
- Remove 15 no-op backward-compat exports from doctor.ts
- Remove getTokenFingerprint no-op from browser/index.ts
- Rename PlaywrightMCP → BrowserBridge across all source files
(backward-compat alias kept in mcp.ts and browser/index.ts)
- Remove unnecessary host_permissions from extension manifest
- Sync extension package.json version to 0.2.0
- All 14 tests pass
All opencli operations now run in a dedicated Chrome window instead
of hijacking the user's active tab. The automation window:
- Created on first command via chrome.windows.create({ focused: false })
- 1280x900 viewport, auto-cleaned up when closed
- All tabs resolved within this window only
- User's main browsing session is never touched
Tested: twitter trending ✅, zhihu hot ✅
Both commands now scroll the conversation list to load more items
before processing. Scrolls up to 20-30 times, stops after 3
consecutive scrolls with no new items loaded.
Previously limited to ~14 visible conversations, now loads as many
as needed (up to --max).
- Rewrite accept.ts: use [data-testid=conversation] click-based approach
instead of extracting href links (requests page has no /messages/xxx links)
- Support comma-separated keywords for OR matching (e.g. '群,微信')
- Add timeoutSeconds: 600 (10 min) for batch DM operations
- Bump default OPENCLI_BROWSER_COMMAND_TIMEOUT from 45s to 60s
- Track visited conversations to avoid infinite loops
Usage:
opencli twitter accept --keyword '微信' --max 20
Workflow:
1. Navigate to /messages/requests
2. Click into each conversation
3. If message contains keyword, click Accept
4. After accept (auto-redirects to /messages), go back to requests
5. Repeat until --max reached or no more matches
process.execPath is always plain 'node' even under tsx,
so .ts files could not be executed. Use --import tsx/esm
flag to enable TypeScript loading in spawned daemon.
Add weread adapter for issue #82, covering search, rankings, book details,
bookshelf, notebooks, highlights, and notes.
Public commands (no login required):
- weread search <keyword> — search books
- weread ranking [category] — book rankings (all/rising/category ID)
Private commands (cookie auth via browser):
- weread book <bookId> — book details
- weread shelf — personal bookshelf
- weread notebooks — books with highlights/notes
- weread highlights <bookId> — underlines in a book
- weread notes <bookId> — personal notes on a book
Closes#82
goto() triggers a full page navigation that resets the JS execution
context, wiping any previously injected fetch/XHR monkey-patches.
The old code installed the interceptor on x.com then navigated away,
so the interceptor was always destroyed before it could capture data.
Fix: navigate directly to the target page, install interceptor after
page load, then scroll to trigger API calls via pagination.
Also fixes the same bug in notifications.ts.
Closes#86
jackwener