* refactor(browser): replace --session flag with <sessionname> positional
The `--session <name>` flag was semantically required but syntactically
optional, which is an anti-pattern. Required + flag is a contradiction:
flag form implies "optional", required is a runtime patch on top. Session
is OpenCLI's "operation target" identifier — the natural form for that is
a positional argument, like `docker exec <container> <cmd>` or
`git checkout <branch>`.
New surface:
opencli browser <sessionname> open https://x.com
opencli browser <sessionname> click 12
opencli browser <sessionname> bind
opencli browser <sessionname> unbind
Commander 14 cannot natively combine a parent positional with subcommand
dispatch — the parent's positional is shadowed by subcommand matching. To
bridge that, main.ts now pre-processes argv: when the token after `browser`
is non-flag and not a known subcommand name, it is treated as the
sessionname and rewritten to the internal `--session <name>` flag form
before commander parses it. Help text on the `browser` command is
overridden via `.usage('<sessionname> <command> [options]')` so users see
the positional form.
Reserved subcommand names (33) are listed in cli-argv-preprocess.ts and
tested for parity with cli.ts subcommand registrations. If a future
subcommand is added, the test fails loudly.
Synced surfaces:
- README.md / README.zh-CN.md — all examples
- docs/guide/browser-bridge.md (+ zh)
- skills/opencli-browser/SKILL.md (bind/unbind, examples, table)
- skills/opencli-usage/SKILL.md
- tests/e2e/browser-tabs.test.ts
- CHANGELOG.md (Unreleased BREAKING)
The internal `--session` flag and the unit tests calling
`program.parseAsync(['...', 'browser', '--session', 'foo', ...])` are
preserved as a stable internal API: tests bypass main.ts pre-processing
and exercise commander directly. The pre-processor has its own targeted
test file (cli-argv-preprocess.test.ts, 10 tests, all green).
Verification:
- npx tsc --noEmit — pass
- npx vitest run --project unit — 1073/1074 pass (1 unrelated skip)
- npx vitest run --project extension — 61/61 pass
- npm run check:typed-error-lint — baseline 189
- npm run check:silent-column-drop — baseline 103
* fix(cli-argv): only rewrite when `browser` is the root command
The preprocessor was looping through every argv slot and would mis-rewrite
occurrences of the literal word `browser` deeper in argv (e.g. `opencli
adapter init browser/x` or arg values containing `browser`).
Now the preprocessor walks past leading root flags + their values to
identify the root command token, and only acts when that token is
`browser`. The full set of root value-consuming flags
(`ROOT_VALUE_FLAGS`) is documented inline and kept in sync with the
`program.option()` calls in cli.ts.
Adds regression tests:
- `opencli adapter init browser x` not rewritten
- URL/path values containing `browser` not rewritten
- `list browser state` (different root command) not rewritten
- `--profile work browser foo state` correctly identifies `foo` as
sessionname (not as --profile's value)
- `--profile=work` long-form-with-equals consumes one slot only
- boolean flags (`-v`) don't consume the next value
12/12 preprocessor tests pass.
* fix(cli-argv): hide --session flag, fail-fast on retired form, rename to <session>
Three blockers in #1505 review:
1. `--session` flag was still visible in `opencli browser --help` and could
be used as a public entrance, contradicting "positional only" UX.
Fix: switch from `.requiredOption()` to `.addOption(new Option(...).hideHelp())`.
The flag is preserved as an internal API for the daemon protocol and direct
`program.parseAsync` callers (tests), but is no longer documented or
surfaced in structured help.
2. `opencli browser --session foo state` still succeeded. Now the argv
preprocessor throws `BrowserSessionArgvError` when root `browser` is
followed by `--session`, and main.ts catches it and exits with a
user-facing usage error pointing to the positional form.
3. Missing-session error message exposed the internal flag:
`required option '--session <name>' not specified`. Now `getBrowserSession()`
in the action body throws `<session> is a required positional argument:
opencli browser <session> <command>`, and commander no longer guards the
hidden option.
Also (per @WAWQAQ) rename placeholder `<sessionname>` -> `<session>` everywhere
user-facing — shorter, matches CLI convention. The help text "<session> is a
required positional: pass the name of the browser session..." carries the
"name" semantics in description, not in the placeholder itself.
Sync surfaces:
- src/cli.ts — usage line, addOption with hideHelp, descriptions
- src/cli-argv-preprocess.ts — throw on --session form
- src/cli-argv-preprocess.test.ts — refusal test for old form
- src/cli.test.ts — assertions updated for hidden option + new error path
- src/help.ts — read `_usage` private field to respect `.usage()` override
(commander's `.usage()` getter returns auto-generated form if not set,
which would otherwise pollute every namespace's usage string)
- src/main.ts — catch BrowserSessionArgvError, stderr + exit
- README.md / README.zh-CN.md
- docs/guide/browser-bridge.md / docs/zh/guide/browser-bridge.md
- skills/opencli-browser/SKILL.md / skills/opencli-usage/SKILL.md
- CHANGELOG.md
Manual smoke tests (against built dist):
- `opencli browser --help` shows `Usage: opencli browser <session> <command> [options]`
- `opencli browser --help` Options block does NOT show `--session`
- `opencli browser --session foo state` → friendly error, no commander stacktrace
- `opencli browser state` → `<session> is a required positional argument: opencli browser <session> <command>`
- `opencli browser foo state` → parses correctly
* fix: inject <session> into subcommand help paths and drop stale sessions ref
Two follow-up blockers from #1505 review:
1. Subcommand help and structured help still rendered the command path
without the parent's positional. `opencli browser foo state --help`
showed `Usage: opencli browser state [options]`, which would lead
users (and agents reading structured help) to think
`opencli browser state` was a valid invocation. Now:
- `commanderPath()` injects an ancestor's leading-positional placeholder
(extracted from its `.usage()` override) between the ancestor's name
and the next path segment when building paths upward.
- `commandPathFromRoot()` strips placeholder segments (e.g. `<session>`)
from the relative `name` field so agents can still address subcommands
by their leaf name; placeholders remain in the `command` / `usage`
display paths.
- `program.configureHelp({ commandUsage: ... })` is applied recursively
to every descendant of `browser`, because commander does NOT inherit
`configureHelp` into subcommands.
Result:
opencli browser <session> click --help
-> Usage: opencli browser <session> click [target] [options]
Daemon, plugin, adapter, profile namespaces (no `.usage()` override)
are unaffected.
2. `skills/opencli-browser/SKILL.md` still referenced
`opencli browser sessions`, which was removed in #1470. Replaced the
sentence with the underlying invariant ("Bound sessions have no
OpenCLI idle-close timer; the binding lasts until `unbind`, tab close,
window close, or daemon restart") without mentioning the deleted
command.
Tests:
- cli.test.ts: structured help expectations updated to include
`<session>` in command/usage paths (3 tests)
- cli-argv-preprocess.test.ts: 12 tests still green
- 1136/1137 unit+extension green (1 unrelated skip)
- typed-error-lint baseline 189
- silent-column-drop baseline 103
* feat(rednote): add rednote.com adapter mirroring xiaohongshu read commands (#1136)
Implements rednote.com support as discussed in issue #1136. The mainland
xiaohongshu adapter stays in place; international users redirected to
www.rednote.com now have a CLI without a copy-pasted adapter.
Issue #1136 documents that xiaohongshu and rednote share DOM selectors,
URL paths, API paths, response schema, cookies, and the xsec_token auth
mechanism. The only material differences:
Layer xiaohongshu rednote
Web host www.xiaohongshu.com www.rednote.com
API host edith.xiaohongshu.com webapi.rednote.com
Security host fe-static.xhscdn.com as.rednote.com
Cookie root .xiaohongshu.com .rednote.com
Search gate Inline text Full-screen modal + text
## Architecture (minimal)
`clis/xiaohongshu/*` keep all selector / regex / extraction logic. Each
command file is touched minimally to export the IIFE or pipeline so the
sibling adapter can reuse it:
search.js + export const buildSearchExtractJs(webHost)
+ export const command = cli({...})
note.js + export const NOTE_EXTRACT_JS
+ export const command = cli({...})
comments.js + export function buildCommentsExtractJs(withReplies)
+ export parseCommentLimit
+ export const command = cli({...})
download.js + export function buildDownloadExtractJs(noteId)
(CDN allowlist now includes rednote alongside xhscdn)
+ export const command = cli({...})
user.js + export const USER_SNAPSHOT_JS
+ export const command = cli({...})
feed.js + export function buildFeedPipeline(webHost)
+ export const command = cli({...})
notifications.js + export function buildNotificationsPipeline(webHost)
+ export const command = cli({...})
note-helpers.js buildNoteUrl now accepts `cookieRoot` + `signedUrlHint`
options (defaults preserved so xhs callers and tests
are unchanged)
user-helpers.js buildXhsNoteUrl / extractXhsUserNotes accept an
optional `webHost` argument (default xhs)
The `export const command = cli({...})` pattern matches twitter/lists.js
and clis/discord-app/*; without it the build-manifest scanner attributes
xhs's command to whichever rednote sibling triggered the transitive
import first.
## clis/rednote/ — thin shims
Each rednote command file imports the relevant builder / constant from
its xiaohongshu sibling and calls `cli()` with the rednote host triple.
No selectors, regexes, or extraction logic are duplicated.
search.js imports buildSearchExtractJs + noteIdToDate
declares its own WAIT_FOR_CONTENT_JS (modal + text
login-gate variants — the one xhs behaviour that
genuinely differs)
note.js imports NOTE_EXTRACT_JS + buildNoteUrl + parseNoteId
comments.js imports buildCommentsExtractJs + parseCommentLimit
+ buildNoteUrl + parseNoteId
download.js imports buildDownloadExtractJs + buildNoteUrl + parseNoteId
user.js imports USER_SNAPSHOT_JS + extractXhsUserNotes
+ normalizeXhsUserId
## Scope (initial)
Ships the five commands verified live against the user's logged-in
rednote.com session: search / note / comments / user / download.
`feed` and `notifications` are intentionally left out. Both rely on
intercepting the xiaohongshu Pinia store at the `homefeed` / `you`
capture pattern; live verification on rednote returns `tap → dict
(error)` for the feed step, so shipping them would surface a broken
contract. The mainland xiaohongshu commands continue to work. Adding
the rednote-side feed / notifications is straightforward follow-up
work once someone with rednote access maps the network surface.
Creator-center commands (publish, creator-*) have no rednote
counterpart and stay xiaohongshu-only, per the reporter's note in #1136.
## Verification
- clis/xiaohongshu/ + clis/rednote/: 103/103 tests green
- npx tsc --noEmit: clean
- npm run build: 807 manifest entries (xhs 13 + rednote 5 + everything
else preserved)
- silent-column-drop / typed-error-lint: 103 / 189 baseline entries,
no new violations
- Live verify against the user's rednote.com session:
rednote search "travel" --limit 1 → real note row
rednote note <signed-url> → 7 field/value rows
rednote comments <signed-url> --limit 3 → 3 top-level rows
rednote user 5b21f6564eacab3b38f05c39 --limit 2 → 2 profile notes
Spaced 15–30s between runs per the xhs/rednote rate-limit guidance;
no write commands invoked. Regression check: xiaohongshu/feed on
the existing mainland session still returns the standard 6-field
rows after the refactor.
Closes#1136
* fix(rednote): tighten adapter failure boundaries
---------
Co-authored-by: jackwener <jakevingoo@gmail.com>
Follow-up to feat #1458 (registering tg-cli/discord-cli/wx-cli in
src/external-clis.yaml) — README and README.zh-CN had not been updated
to reflect the new entries.
Updates four spots in each README:
- intro paragraph that names example external CLIs
- "CLI Hub" highlight bullet
- "OpenCLI is not only for websites" bullet list
- the External CLI table itself
Move the Browser Bridge daemon WebSocket out of the MV3 service worker and into an offscreen document. Remove the popup/action UI and obsolete extension log forwarding now that doctor is the diagnostic surface.
* feat(claude): add Claude adapter
Adds a Claude (claude.ai) browser adapter family with seven commands
modeled on the existing clis/deepseek/ pattern: ask, send, new, status,
read, history, detail.
Closes#1251
* feat(claude): align send command columns with doubao
Match the established Status / SubmittedBy / InjectedText shape used by
doubao send so agent loops can rely on a consistent fire-and-forget
output across AI chat adapters.
* fix(claude): preserve DOM order in getVisibleMessages
The previous implementation queried user-message and assistant-message
nodes in two passes, which serialized as [u1, u2, u3, a1, a2, a3] for
multi-turn chats instead of the correct conversation order. Single
combined query preserves DOM order so claude read / detail return
turns in the order the user reads them on the page.
* docs(claude): note --live requirement for read across invocations
* fix(claude): fail fast on auth and empty states
---------
Co-authored-by: jackwener <jakevingoo@gmail.com>
* feat(google-scholar): add cite and profile commands, fix search dedup
- cite: get BibTeX/EndNote/RefMan/RefWorks citation for a paper.
Clicks the cite button in search results and fetches the citation
content from Google's citation endpoint.
- profile: view an author's Google Scholar profile (h-index,
i10-index, citation count, top papers). Accepts author name
or Scholar user ID.
- search: fix duplicate results caused by CSS selector matching
both outer container (.gs_r.gs_or.gs_scl) and inner child
(.gs_ri) for each paper.
Closes#1174, closes#1175
* fix(google-scholar): fail fast on cite and profile misses
* fix(google-scholar): document new commands and lock dedup test
---------
Co-authored-by: jackwener <jakevingoo@gmail.com>
* docs: update extension install to Chrome Web Store link
Extension is now published on Chrome Web Store. Replace manual
download/unpack instructions with the store link across READMEs
and skill docs.
* docs: restore manual install as Option B alongside Chrome Web Store
--live (OPENCLI_LIVE=1) keeps the automation window open after an adapter
command finishes, so agents or humans can inspect the page state. Default
behavior (immediate closeWindow) is unchanged.
--focus (OPENCLI_WINDOW_FOCUSED=1) surfaces the existing env-var toggle as a
CLI flag so users don't need to shell-export to see the window in foreground.
Both flags are parsed early in main.ts and stripped from argv, so they can be
placed anywhere on the command line and work on any subcommand (adapter or
browser).
* feat(bilibili): add video command
Add `opencli bilibili video <bvid|url|short-link>` to fetch one
video's metadata via the public /x/web-interface/view endpoint.
Returns title, author, category, publish time, duration, view /
danmaku / reply / like / coin / favorite / share counts, parts,
thumbnail, and description as a key/value table.
Reuses `resolveBvid` and `apiGet` from clis/bilibili/utils.js to
stay consistent with the existing bilibili adapters
(subtitle/search/etc. all follow the same navigate + apiGet
pattern). Non-zero API codes surface as CommandExecutionError.
Fills a visible gap: existing bilibili commands cover search,
hot, subtitle, ranking, user-videos etc., but nothing returned
metadata for a single video — `web read` only gets a DOM shell.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix(manifest): register bilibili video command
---------
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Co-authored-by: jackwener <jakevingoo@gmail.com>
* refactor: consolidate 6 skills into 3, remove mechanical commands
Replaces opencli-oneshot / opencli-explorer / opencli-browser /
opencli-usage with a single opencli-adapter-author skill that takes
the AI agent end-to-end: site recon, API discovery, field decoding,
adapter coding, and `opencli browser verify`.
Removes the mechanical commands (`explore`, `synthesize`, `generate`,
`cascade`, `record`) and their src/tests — they were codegen scaffolding
meant for agents, which the new skill handles more flexibly via
`opencli browser` primitives.
Skill highlights:
- Top-level decision tree + 12-step runbook
- 5 site patterns (SPA / SSR / JSONP / Token / Streaming)
- 5-layer API discovery (network → initial state → bundle → token → interceptor)
- Field decode playbook (self-explanatory → codes → sort-key comparison)
- Output design guide (columns, types, order, ≤15 per adapter)
- Two-layer site memory: in-repo seeds for eastmoney/xueqiu/bilibili/tonghuashun
plus local `~/.opencli/sites/<site>/` runtime workspace
Kept skills: opencli-autofix (now points to adapter-author for rewrites),
smart-search. Kept primitives: `browser *`, `doctor`, `list`, `validate`,
`verify`, `<site> <cmd>`, `plugin *`, `completion`.
No backward compatibility shims. Full test suite (1605 tests) passes.
* review fixes: honest coverage, hard memory-hit path, typo, stale docs
- site-memory hit path no longer jumps to writing adapter; forces Step 5
endpoint re-verification + Step 7 field check, and 30-day expiry
- site-memory.md now specifies exact schemas for endpoints.json /
field-map.json / notes.md / fixtures + write-back timing rules
- coverage-matrix.md marks unverified patterns as 🟡 with an evidence
section citing coingecko dry run + PR #1091 eastmoney + bilibili
- eastmoney seed typo: resolveSecids -> resolveSecid (and splitSymbols)
- docs/developer/ai-workflow.md rewritten to teach the adapter-author
skill + opencli browser * primitives (dropped generate/synthesize/
cascade/explore references)
- ts-adapter.md, getting-started.md, CHANGELOG.md:87 updated to point
at opencli-adapter-author
* fix(ci): resync package-lock + drop stale built-in list reference
- Regenerate package-lock.json to restore @emnapi/core + @emnapi/runtime
entries that got dropped during the rebase — `npm ci` was failing on all
CI jobs (build / audit / docs-build / bun-test / unit-test)
- docs/guide/getting-started.md: built-in list dropped `explore`, now
reads (list, validate, verify, browser, doctor, plugin...)
* fix(ci): restore package-lock.json from main (unrelated lockfile churn)
From first principles, `opencli browser` commands exist for AI Agents to
operate websites through the browser skill. Reframe both READMEs to reflect
this: show users how to install the skill into their AI agent and describe
tasks in natural language, rather than listing raw CLI commands.
* feat(twitter): rewrite lists via GraphQL + add list-tweets
The DOM-scraping / detail-click approach in PR #1053 remained fragile
against X's frequent overview-page rendering changes and slow (N+1 page
loads per list). Rewrite `twitter lists` to call
`ListsManagementPageTimeline` GraphQL directly — one request returns all
owned + subscribed lists with id/name/member_count/subscriber_count/mode.
Also add `twitter list-tweets <listId>` for pulling the tweet stream from
a list, completing the read-side chain (lists → pick an id → list-tweets).
- lists: drop positional `user` arg (GraphQL returns only logged-in
user's lists), add `id` column, change followers to exact integer from
subscriber_count.
- list-tweets: same GraphQL pattern as bookmarks/likes (BEARER + ct0 +
dynamic queryId with static fallback + cursor pagination).
- Delete obsolete lists-parser.js and lists.d.ts.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* feat(twitter): add list-add / list-remove with Save-button commit
Two new commands to toggle list membership. X's list dialog uses a
"click-to-stage, Save-to-commit" pattern — clicking a row only updates
optimistic UI; the actual POST fires when the user clicks the top-right
"Save" button. Pressing ESC or the close-X silently cancels the change.
Implementation:
- Resolve listId → name via ListsManagementPageTimeline GraphQL, so we
match the dialog row by name (dialog rows have no data-testid listId).
- Open profile page → DOM click "…" menu → "Add/remove from Lists".
- Scroll dialog to locate target row (virtualized list).
- page.nativeClick on row — trusted CDP Input.dispatchMouseEvent fires
React's onclick, flips aria-checked (.click() alone does not suffice;
X ignores non-trusted events for list mutations).
- page.nativeClick on the Save button — commits to server.
- Verify by re-fetching ListsManagementPageTimeline and diffing
member_count: success only if N→N±1. No silent successes.
This fixes the pattern where batch `list-add` calls returned success for
every user but committed zero to the server (optimistic UI lied).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
* fix: stabilize twitter list manifest and query ids
* docs: add twitter list command discoverability
---------
Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Co-authored-by: jackwener <jakevingoo@gmail.com>
* fix(xiaoyuzhou): migrate from broken SSR scraping to authenticated API (fixes#1023)
Xiaoyuzhou removed SSR rendering — /podcast/<id> and /episode/<id> pages
now return 404, breaking fetchPageProps() which scraped __NEXT_DATA__.
Migrate podcast, podcast-episodes, episode, and download commands to use
the existing authenticated API client (requestXiaoyuzhouJson) that
transcript.js already uses successfully.
Changes:
- podcast.js: use /v1/podcast/get API endpoint
- podcast-episodes.js: use /v1/podcast/listEpisode API endpoint
- episode.js: use /v1/episode/get API endpoint
- download.js: use /v1/episode/get API endpoint
- utils.js: remove unused fetchPageProps, keep format helpers
- Update all affected tests (download.test.js, utils.test.js)
- Change strategy from PUBLIC to LOCAL (requires credentials)
* fix(xiaoyuzhou): align local strategy contract
* fix(xiaoyuzhou): align local api metadata
---------
Co-authored-by: jackwener <jakevingoo@gmail.com>
* fix: remove duplicate extension zip from releases
The release and build-extension workflows were creating both
opencli-extension.zip and opencli-extension-v{version}.zip (identical
content), causing both to be uploaded. Keep only the versioned filename.
* docs: update extension zip filename to versioned format
Update all references from opencli-extension.zip to
opencli-extension-v{version}.zip to match the workflow change.
Add missing env vars to both README and README.zh-CN:
- OPENCLI_SKIP_FETCH: skip adapter sync on global install
- OUTPUT: override output format (json/yaml/table)
- DEBUG=opencli: internal debug logging
- DEBUG_SNAPSHOT: DOM snapshot debug output
* feat(twitter): add lists command to retrieve user lists
Add twitter/lists command that fetches Twitter/X lists for a user.
Supports:
- Lists with member and follower counts
- Private/public mode detection
- Default to current user if no user specified
- Works for any Twitter user
* docs: add lists command to twitter commands in README
Add twitter lists command to Built-in Commands table in both
English and Chinese README files
* fix(twitter): parse lists from card DOM instead of locale-specific page text
---------
Co-authored-by: isanwenyu <isanwenyu@users.noreply.github.com>
Co-authored-by: jackwener <jakevingoo@gmail.com>
* fix: project hygiene — docs, lint, daemon restart, code fence
- Update Node version requirement from >= 20 to >= 21 in 7 doc files
(README, README.zh-CN, installation guides, troubleshooting)
- Update adapter count from 79+ to 87+ in READMEs
- Remove duplicate `lint` script (identical to `typecheck`)
- Fix TESTING.md CI matrix: Node ['22'] instead of ['20', '22']
- Fix autofix SKILL.md code fence escaping (\``` → ~~~)
- Add daemon restart to postinstall so updated adapters are picked up
- Fix preuninstall to respect OPENCLI_DAEMON_PORT env var
* fix: align docs and skills with JS-first adapter contract
Adapters are now .js files (not .ts). Update all references across:
- README.md, README.zh-CN.md, CONTRIBUTING.md
- docs/guide/getting-started.md, docs/index.md
- skills/opencli-browser/SKILL.md, skills/opencli-explorer/SKILL.md
The runtime (discovery.ts) only loads .js from user clis/ directories,
and `opencli browser init` generates .js scaffolds. Documentation was
still teaching users to create .ts files.
* fix: update CI matrix to Node 22 only (drop Node 20)
package.json requires Node >= 21 (styleText dependency). The CI matrix
was still testing Node 20 which doesn't meet this requirement.
* fix: revert incorrect daemon restart from postinstall
The daemon (browser bridge) only handles CDP communication — it has no
knowledge of adapters. Adapter discovery, loading, and execution all
happen in the CLI process, which is fresh each invocation. The
_loadedModules cache in execution.ts is process-local and not a real
staleness concern. Remove the unnecessary restartDaemon() call.
* feat: auto-close adapter windows, add OPENCLI_WINDOW_FOCUSED, document config
1. Adapter commands now close the automation window immediately after
completion instead of waiting for the 30s idle timeout.
2. OPENCLI_WINDOW_FOCUSED=1 opens automation windows in the foreground
(useful for debugging). Default remains background.
3. Add Configuration section to README (EN/ZH) and opencli-usage skill
listing all stable user-facing environment variables.
* Fix OPENCLI_WINDOW_FOCUSED to be per-request, not frozen at daemon startup
Move env var read from daemon (startup-time constant) to CLI side
(sendCommandRaw), so it works correctly with the persistent daemon model.
Each request now reads the env var fresh and includes windowFocused in
the command payload.
jakevin