git add fails with fatal error when a pathspec glob doesn't match
any files (e.g. 'extension/src/**/*.ts' in repos without that dir).
Try each glob individually and skip non-matching ones silently.
Fixes regression from #826.
- autoresearch/debug.ts: prompt via stdin instead of shell interpolation
- autoresearch/engine.ts: git commit via execFileSync instead of shell
string with quote escaping
- autoresearch/eval-skill.ts: use execFileSync with argument array,
pass prompt via stdin
- src/plugin.ts: use execFileSync for esbuild PATH lookup
Same shell injection pattern as fix.ts — prompt with $, backticks,
or other metacharacters would be expanded by the shell when
interpolated into the command string.
- spotify/spotify.ts: replace shell-interpolated `open "${url}"` with
execFileSync to prevent URL-based shell injection
- chatgpt/ax.ts: replace execSync shell strings with execFileSync for
osascript calls (activate + delay)
- chatgpt/ask.ts: replace execSync `sleep` with spawnSync to avoid
shell interpolation
- record.ts: fix misleading "idempotent injection" comment — the
function is safe to call multiple times but not strictly idempotent
- engine.ts: replace `git add -A` with scope-aware `execFileSync` to
stage only files matching config.scope globs, and guard against empty
scope degenerating into staging all files
- fix.ts: pass prompt via stdin `input` option instead of shell string
interpolation to prevent $, backtick, and other metacharacter expansion
- generate.ts: update stale comment that claimed unimplemented pipeline
steps (register, verify, Strategy Cascade)
* refactor: remove scoring heuristic, replace with noise filter + metadata
The scoring mechanism was a pre-LLM heuristic that compressed rich endpoint
metadata into a single number. Since this project is designed for AI Agents,
the agent can reason about structured metadata directly.
Changes:
- Remove scoreEndpoint/scoreRequest/scoreWriteRequest and all score fields
- Replace with isNoiseUrl() filter (tracking/beacon/pixel) + isUsefulEndpoint()
- Remove artificial confidence percentages (was score/20)
- Sort by itemCount (transparent, observable) instead of weighted score
- Endpoints now expose full structured metadata for agent consumption
- Net reduction: -43 lines
* fix: widen endpoint filter to keep single-object JSON and stats/metric URLs
- Remove stats/metric from noise pattern — these are often business APIs
- Relax isUsefulEndpoint to keep any JSON endpoint, not just arrays
(preserves /me, /profile, /detail and other single-object APIs)
* fix: add deterministic endpoint ordering for generate/synthesize path
The AI agent path doesn't need ranking, but generate/synthesize still
pick candidates[0] as default — this needs a stable, explainable order.
- Add endpointSortKey() with transparent observable signals: array items,
detected fields, API path patterns, query params
- Update synthesize chooseEndpoint fallback to use itemCount + field count
- Sort key is internal only; not exposed as score to external consumers
* feat(linux-do): split topic content into a dedicated command
Move the old main-post path out of linux-do topic so topic stays a summarized first-page reader while topic-content becomes the Markdown-focused entrypoint for full post bodies.
Ultraworked with [Sisyphus](https://github.com/code-yeongyu/oh-my-openagent)
Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>
* feat(linux-do): update topic content handling to include YAML front matter
* fix(linux-do): update default output format to plain for topic-content rendering
* fix(linux-do): replace js-yaml with inline YAML serialization for topic-content
Adapters must only import node builtins, relative modules, or opencli
public APIs. Hand-roll the simple front matter serialization to remove
the third-party js-yaml dependency.
* fix(linux-do): refine YAML quoting to only escape colons followed by space
Colons in URLs (e.g. https://) are valid unquoted YAML values. Only
quote when a colon is followed by a space or appears at end of line.
---------
Co-authored-by: Sisyphus <clio-agent@sisyphuslabs.ai>
* refactor: extract shared scoring logic and consolidate time format utils
- Extract applyUrlScoreAdjustments() and scoreArrayResponse() to analysis.ts,
eliminating duplicated endpoint scoring between explore.ts and record.ts
- Consolidate formatDuration/formatUptime into a single formatDuration(ms)
in download/progress.ts, reused by commands/daemon.ts
* fix: preserve explore scoring semantics and round daemon uptime
- Revert explore.ts scoreEndpoint to original inline /api/ /x/ bonus
without record's tracking/analytics penalty (blocker from review)
- Math.round uptime*1000 to avoid floating-point noise in daemon status
* feat(xueqiu): add kline and groups adapters
Add kline.yaml: fetch candlestick/OHLCV data from Xueqiu v5 chart API.
Supports custom days lookback and outputs date, open, high, low, close,
volume, percent.
Add groups.yaml: list Xueqiu portfolio/group entries.
* fix(xueqiu): correct groups.yaml to use /portfolio/list.json API
The previous implementation used /portfolio/stock/list.json which only
returns stocks in a single group and does not return the group list.
Switch to /portfolio/list.json which returns all portfolio groups
including 实盘, 沪深, 港股, 美股, 模拟(pid=-4), 持仓 etc.
* fix(xueqiu): replace watchlist category param with pid selector
- Remove the unused 'category' parameter (the API ignores it;
all groups live under category=1 regardless)
- Replace with 'pid' parameter to allow fetching any group:
-4=simulated, -5=SH/SZ, -6=US stocks, -7=HK stocks, etc.
- API path still uses category=1 but pid is now user-controllable
---------
* feat(operate): unify network capture + implement CDP consoleMessages
- operate open: start session capture before navigation (catches initial requests)
- operate network: prefer readNetworkCapture() over JS interceptor
- CDPPage: implement consoleMessages() via Runtime.consoleAPICalled
Part of #810
* fix(operate): use correct daemon/CDP entry field names for network capture
Daemon and CDP capture entries use responseStatus/responseContentType/
responsePreview (not status/contentType/responseBody). Fix the
normalization in operate network to match the actual entry shape from
extension/src/cdp.ts.
* fix(cdp): capture Runtime.exceptionThrown in consoleMessages
- Register Runtime.exceptionThrown handler to capture uncaught exceptions
as error-level messages (most valuable diagnostic signal)
- 'error' filter now returns both console.error() and warning/exception
entries, matching typical severity-based logging semantics
* feat(cdp): implement session-level network capture for CDPPage
Implements startNetworkCapture() and readNetworkCapture() on CDPPage using
CDP Network domain events. Updates explore.ts to prefer session capture
over Performance API networkRequests().
Closes part of #810
* fix(cdp): use Network.loadingFinished for reliable body capture
- Move getResponseBody call from responseReceived to loadingFinished,
matching the extension's implementation pattern
- Use extension-compatible entry shape (responseStatus, responseContentType,
responsePreview) instead of custom field names
- Remove unreliable 100ms sleep hack in readNetworkCapture()
- Align with extension/src/cdp.ts:419-437 for consistency
* fix(cdp): drain buffer on readNetworkCapture to match daemon contract
readNetworkCapture() must clear the buffer after reading, matching the
daemon Page's read-and-drain behavior. Without this, repeated reads
would return stale entries.
* fix(cdp): await in-flight body fetches before returning from readNetworkCapture
Track all pending getResponseBody promises and await them in
readNetworkCapture() before draining the buffer. This ensures
explore/diagnostic consumers always get entries with responsePreview
populated, not empty shells where the body fetch hasn't resolved yet.
* fix(explore): handle both legacy and capture entry field names
parseNetworkRequests now maps both shapes:
- Legacy: status, contentType, responseBody
- Capture (extension/CDP): responseStatus, responseContentType, responsePreview
Also clears _pendingBodyFetches on startNetworkCapture reset.
* fix: add safety boundaries to diagnostic output
- Redact sensitive headers (Authorization, Cookie, etc.) from network requests
- Redact sensitive URL query parameters (token, key, secret, etc.)
- Cap individual fields: snapshot (100K chars), adapter source (50K chars),
network requests (50 entries, 4K body each), stack trace (5K chars)
- Enforce 256KB total output budget with graceful degradation:
drops snapshot first, then page state entirely
- Export truncate/redactUrl helpers for testing
* fix: add free-text redaction for all diagnostic string channels
Addresses review feedback: snapshot, consoleErrors, error message/hint/stack
could contain inline secrets (Bearer tokens, JWTs, cookie values, token=value
patterns). All string channels now pass through redactText() before emission.
- Add redactText() with patterns for Bearer tokens, JWTs, cookie values,
and inline key=value secrets
- Apply redactText to: error.message, error.hint, error.stack,
page.snapshot, page.consoleErrors
- Add 6 new test cases for redactText and error message redaction
* fix: resolve adapter source path and add page state collection timeout
Fixes#808 items 1 and 3:
1. adapter.source was missing for all command types because buildRepairContext
only checked cmd._modulePath (set only for manifest lazy-loaded TS).
Now resolveAdapterSourcePath() checks cmd.source first, skips manifest:
pseudo-paths, and maps dist/clis/*.js back to source clis/*.ts.
3. collectPageState() had no timeout — a hung CDP connection would block
error propagation indefinitely. Now wrapped with 5s Promise.race timeout,
falling back to emitting diagnostic without page state.
* fix: track sourceFile in manifest for YAML adapter source resolution
YAML commands inlined in the manifest previously lost their original file
path, causing resolveAdapterSourcePath() to return undefined. Add
sourceFile field to ManifestEntry so discovery can reconstruct the
editable source path for both YAML and TS commands.
* feat: add structured diagnostic output for AI-driven adapter repair
When OPENCLI_DIAGNOSTIC=1 is set, failed commands emit a RepairContext
JSON to stderr containing the error, adapter source, and browser state
(DOM snapshot, network requests, console errors). AI Agents consume
this to diagnose and fix adapters when websites change.
Also adds the opencli-repair skill guide for AI Agents.
* fix: correct e2e test binary path to dist/src/main.js
The e2e helpers pointed to dist/main.js but the actual build output
is at dist/src/main.js (matching package.json "main" field). This
caused all e2e-headed tests to fail with "Cannot find module".
* fix: correct dist/main.js path in autoresearch scripts
* fix: emit diagnostic for pre-session browser failures
When browser connection fails before the session callback runs
(e.g., BrowserConnectError), the inner diagnostic catch never fires.
Use a flag to ensure the outer catch emits diagnostic as a fallback.
* test: tolerate unavailable Bloomberg RSS feeds in e2e
* test: skip flaky bloomberg businessweek e2e test
The Bloomberg Businessweek RSS feed is intermittently unavailable,
causing CI failures unrelated to code changes.
* revert: restore bloomberg businessweek e2e coverage
* fix: avoid inserting completion config inside multi-line shell commands
The postinstall zshrc insertion logic splits backslash-continued blocks
(e.g. zinit stanzas) when it finds a compinit match inside them, which
breaks the user's shell config. Walk backward past continuation lines
so the insertion lands before the entire logical command.
* fix: append zsh completion to end of .zshrc instead of splicing
Replace the fragile compinit-searching splice logic with a simple
append, matching the strategy already used for bash. This avoids
breaking multi-line commands (e.g. zinit blocks with zicompinit).
Still detects existing compinit to avoid adding a duplicate call.
* fix: stop modifying shell rc files in postinstall
Replace the fragile .zshrc/.bashrc modification logic with a safer
approach: only write completion files and print setup instructions.
The previous approach tried to parse and splice into rc files, which
broke multi-line shell commands (e.g. zinit blocks with backslash
continuations matching /compinit/). Instead of attempting to fix the
parser, remove rc modification entirely — this matches the approach
used by rustup, homebrew, and other CLI tools.
Closes#788
---------
Co-authored-by: jackwener <jakevingoo@gmail.com>
* fix(discovery): expose runtime deps to user adapters
* fix: route copied adapters through opencli exports
* refactor: route adapter status output through logger
* refactor: migrate adapter imports to package exports
Replace all relative imports (../../src/registry.js, ../../browser/cdp.js, etc.)
with package exports (@jackwener/opencli/registry, @jackwener/opencli/errors, etc.)
across all 484 adapter files.
This decouples adapter import resolution from directory structure:
- User CLIs in ~/.opencli/clis/ resolve via node_modules symlink
- Internal adapters resolve via Node.js self-referencing
- No more shim files needed for import resolution
Changes:
- package.json: add sub-path exports for all public modules
- clis/**: replace relative imports with @jackwener/opencli/...
- discovery.ts: simplify ensureUserCliCompatShims to symlink-only
- registry-api.ts: export CommandArgs type
- Remove root-level shim directories (browser/, download/, pipeline/)
- Remove shim entries from tsconfig.json include and package.json files
* test: add regression tests for package exports
Prevents regressions like #788/#791 by:
1. Scanning all adapter files for forbidden relative imports
(../../src/, ../../browser/, etc.) — fails if any remain
2. Verifying every package.json export maps to an existing source file
18 new test cases.
* fix: use junction on Windows + broaden test patterns
- discovery.ts: use 'junction' symlink type on Windows (no admin required)
- package-exports.test.ts: generalize forbidden patterns to catch any
depth of ../ traversal (not just ../../ and ../../../)
* fix: update stale vi.mock/importActual paths in adapter tests
Test files still used old relative paths for vi.mock() and
vi.importActual() calls. Updated 5 test files to use package exports.
Also broadened regression test patterns to catch mock/importActual paths.
* fix: use rm instead of unlink for symlink cleanup, add warn on failure
Addresses review feedback from Astro-Han:
- rm() handles both symlinks and stale directories (unlink fails on dirs)
- Log a warning when symlink creation fails instead of silent catch
* docs: update import examples to use package exports
Update all documentation, contributing guides, and skills to use
@jackwener/opencli/registry instead of ../../src/registry.js.
Without this, users following the docs would write adapters with
broken imports since the old shim files are no longer created.
Bug 1: version.ts used a single-level parent lookup for package.json,
which broke after #784 changed rootDir from "src" to "." (version.js
now lives in dist/src/ instead of dist/). Walk up until package.json
is found — works in both dev (src/) and prod (dist/src/).
Bug 2: adapters copied to ~/.opencli/clis/ import ../../src/registry.js
etc., which resolves to ~/.opencli/src/. Derive src/ compat shims from
the existing rootShims list so these imports resolve correctly.
- Replace 140 instances of `src/clis/` → `clis/` across 12 doc files
(path changed after repo restructure)
- Remove non-command `rpc` and `rankings` from notebooklm/amazon
command lists in README, README.zh-CN, SKILL.md, and adapters index
(these are internal utility modules, not user-facing commands)
- Add `deep-research` and `deep-research-result` to gemini adapter doc
- Add `movers-shakers` and `new-releases` to amazon adapter doc
- Update notebooklm doc examples to use canonical commands instead of
deprecated aliases (`metadata` → `get`, `notes-list` → `note-list`)
- Bump SKILL.md version to 1.6.3
* docs: fix outdated commands and adapter counts in README and skills
- Add gemini deep-research and deep-research-result commands
- Fix notebooklm: remove non-existent select/metadata/notes-list, add rpc
- Add amazon movers-shakers, new-releases, rankings commands
- Add missing weibo commands in zh-CN README
- Fix linux-do missing hot/latest/category in zh-CN README
- Update adapter count from 73+ to 79+
- Update skills version to 1.6.2
- Add full spotify command list in skills SKILL.md
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* fix(test): update xiaohongshu note tests for search_result URL change
buildNoteUrl now uses /search_result/<id> instead of /explore/<id> for
bare note IDs. Update test expectations to match:
- buildNoteUrl test: expect /search_result/ not /explore/
- goto URL assertion: expect /search_result/ not /explore/
- empty shell hint: match actual error message text
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* fix(test): update xiaohongshu comments test for search_result URL change
- Bare note ID now navigates to /search_result/ not /explore/
- Full URL inputs are preserved as-is (including /explore/ URLs)
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* refactor: simplify core modules — remove root shims, consolidate error classification, streamline cascade/interceptor, clean up synthesize
1. Remove root-level shim files (errors.ts, logger.ts, registry.ts, types.ts, utils.ts, launcher.ts) — update all ~840 adapter imports to reference src/ directly
2. Consolidate interceptor: reuse shared DISGUISE_FN in tap interceptor instead of reimplementing
3. Unify error classification: single ClassifiedError type with icon/exitCode/hint lookup table, eliminating duplicated pattern matching between resolveExitCode and renderError
4. Simplify cascade probe: replace repetitive switch cases with PROBE_OPTIONS lookup map
12. Clean up synthesize.ts: remove deprecated snake_case field aliases (recommended_args, recommended_columns, recommendedColumnsLegacy) and unnecessary constant aliases
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* docs: update import paths in contributor docs and skill templates
Update all documentation and skill files to reference src/ directly,
matching the shim removal in the previous commit.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* fix: update new Gemini adapter imports to use src/ paths
Fix imports in newly added deep-research adapter files that were
still referencing the deleted root shim files.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* fix: update xiaohongshu tests for /search_result/ URL change
Tests now expect /search_result/<id> for bare note IDs (matching
the note-helpers.ts change from PR #774) and updated empty-shell
hint assertion.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* fix: update LessWrong and hupu adapter imports to use src/ paths
Fix imports in newly merged LessWrong and hupu/mentions adapter
files that were still referencing the deleted root shim files.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* test(xueqiu): mock logger via src path
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* feat(hupu): add hupu cli adapter
* fix(hupu): prevent detail from returning the wrong thread
* refactor: deduplicate shared utilities in hupu adapter
- Merge postHupuJson and postHupuReplyJson into single function with mode parameter
- Move stripHtml and decodeHtmlEntities to utils.ts, remove duplicate definitions
* fix(hupu): add mentions command
* fix: move mentions.ts to clis/hupu/, remove src/clis/hupu duplicates
Post PR #782 restructure: adapter files live at root clis/, not src/clis/.
---------
Co-authored-by: jackwener <jakevingoo@gmail.com>
* feat(lesswrong): add LessWrong adapter
15 commands for the LessWrong rationality and AI alignment community:
- Post listings: curated, frontpage, new, top, top-week/month/year
- Content: read (full post), comments, shortform (quick takes)
- Discovery: tag, tags, sequences
- Users: user (profile), user-posts
All commands use the public GraphQL API (no browser required).
Time-filtered views use the `after` date parameter.
Tag lookup resolves slugs to IDs via the `tagBySlug` view.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix: move lesswrong adapter to clis/ (post PR #782 restructure)
New adapter files were created at src/clis/lesswrong/ but PR #782 moved
all adapters to root clis/. Move to correct location.
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-authored-by: jackwener <jakevingoo@gmail.com>
XHS blocks /explore/<id> without a valid xsec_token, causing code 66
(empty result) when passing bare note IDs. The /search_result/<id> path
works without xsec_token when the user is logged in via cookies.
Changes:
- note-helpers.ts: buildNoteUrl now uses /search_result/<id> for bare IDs
- note.ts: remove isBareNoteId branching and simplify empty shell error
* feat(gemini): add deep-research workflow and docs export result
* fix(gemini): improve deep-research submit and confirm flow
* fix(gemini): return waiting state when deep research is in progress
* fix(gemini): avoid false submit detection on root app transcript changes
* fix(gemini): return pending state when deep-research export is not ready
---------
Co-authored-by: f1480022 <f148002@163.com>
* fix(36kr): replace waitForCapture with DOM polling for search/hot
waitForCapture(6) always times out on 36kr because the API intercept
never captures a matching request. However, the DOM is already fully
rendered with search/hot results by the time the timeout fires.
Replace the 6-second intercept wait with a DOM polling loop that checks
for article links (a[href*="/p/"]) every 300ms, returning immediately
once content is available (typically ~1s vs 6s timeout + error).
Tested on opencli 1.6.2 with both CDP and Browser Bridge modes.
* fix: rebase onto main, remove unused interceptor, fix strategy
- Rebase onto main after clis/ move (PR #782)
- Remove installInterceptor calls (no longer used after waitForCapture removal)
- Change strategy from INTERCEPT to PUBLIC (browser: true) to match actual behavior
- Improve polling loop readability
---------
Co-authored-by: buruguo <buruguo@lambdafintech.com>
Co-authored-by: jackwener <jakevingoo@gmail.com>
* fix: review follow-ups — better first-run log, OPENCLI_FETCH=1 skips version check
- Clarify first-run log message: "copying adapters (one-time setup)"
- Add comment explaining why scriptPath uses two levels of ../
- OPENCLI_FETCH=1 now bypasses version-skip to allow forced refresh
* fix: update doc-coverage script path after clis/ move
check-doc-coverage.sh still referenced src/clis/ after PR #782 moved
adapters to root clis/. This caused CI to fail with "0/1 documented".
* fix: resolve package root dynamically for symlink and first-run paths
The symlink at ~/.opencli/node_modules/@jackwener/opencli pointed to
dist/ instead of the package root in prod mode, breaking user TS CLIs
that import from '@jackwener/opencli/registry'.
The first-run scriptPath also resolved incorrectly in dev mode.
Extract findPackageRoot() that walks up to find package.json, fixing
both paths for dev (src/) and prod (dist/src/) layouts.
* refactor: move adapters from src/clis/ to root clis/ for monorepo separation
Separates CLI adapters from the core runtime to prepare for independent
adapter distribution via postinstall fetch.
Key changes:
- Move src/clis/ → clis/ (adapters at repo root)
- Change tsconfig rootDir from "src" to "." so tsc compiles both
- Create root-level shim files (registry.ts, errors.ts, etc.) so adapter
relative imports (../../registry.js) resolve correctly
- Update build-manifest.ts, main.ts paths for new dist/src/ structure
- Expand ensureUserCliCompatShims() to cover all adapter import targets
(types, utils, logger, launcher, browser/*, download/*, pipeline/*)
- Add scripts/fetch-adapters.js postinstall for ~/.opencli/clis/ sync
- Update vitest.config.ts adapter test paths
- Add package.json files field to exclude adapters from npm package
Official adapter files are unconditionally overwritten on update;
user-created files not in the manifest are preserved.
* fix: add dist/clis/ and cli-manifest.json to npm files, harden fetch-adapters
- Add dist/clis/ and dist/cli-manifest.json to package.json files field
so built-in adapters and manifest ship with the npm package
- Replace execSync with execFileSync to prevent command injection
- Add version check to skip redundant adapter fetches
- Track tmpRoot explicitly for reliable cleanup
* fix: address review blockers — manifest-based updates, global-only fetch, first-run fallback
1. Manifest-based update strategy:
- Read old manifest to identify previously-official files
- Clean up files removed upstream (in old manifest but not new)
- User-created files (never in any manifest) remain untouched
2. Only run fetch-adapters on global install (npm_config_global=true)
or explicit OPENCLI_FETCH=1, preventing heavy side effects for
local/dev installs
3. First-run fallback in discovery.ts:
- ensureUserAdapters() checks for adapter-manifest.json
- If missing and ~/.opencli/clis/ is empty, spawns fetch-adapters.js
- Guarantees adapters are available even with --ignore-scripts
* fix: remove OPENCLI_FETCH env var, use internal _OPENCLI_FIRST_RUN instead
* feat: also support OPENCLI_FETCH=1 for explicit adapter fetch trigger
* simplify: replace git clone with local copy from dist/clis/
Adapters already ship in the npm package (dist/clis/), so there's no
need to clone from GitHub. Copy directly from the installed package:
- Eliminates git, curl, tar dependencies
- No network calls in postinstall
- No timeout/offline issues
- Version always matches the installed CLI
- ~65 lines of clone/download code replaced by one cpSync loop
* feat(hupu): add hupu cli adapter
* fix(hupu): prevent detail from returning the wrong thread
* refactor: deduplicate shared utilities in hupu adapter
- Merge postHupuJson and postHupuReplyJson into single function with mode parameter
- Move stripHtml and decodeHtmlEntities to utils.ts, remove duplicate definitions
---------
Co-authored-by: jackwener <jakevingoo@gmail.com>
* feat: 推特新增回复图片能力支持本地路径和网络路径
* fix(twitter/reply): fix image upload fallback, restore execCommand, add size limit
- Fix attachReplyImage fallback: use uploaded flag instead of checking
page.setFileInput existence, so base64 fallback actually runs when
CDP setFileInput throws "Unknown action"
- Restore execCommand('insertText') as primary text input method for
Twitter's Draft.js editor, with paste event as fallback
- Add 20MB size limit for remote image downloads to prevent OOM
- Remove unsafe buttons[0] fallback that could click invisible buttons
* fix(twitter/reply): add local image size check and base64 fallback warning
Local images were not validated for size — a 100MB file would fail only
at upload time. Remote images already had MAX_IMAGE_SIZE_BYTES checks.
Also add a console.warn when using the base64 fallback with large
payloads, consistent with xiaohongshu/publish.ts behavior.
---------
Co-authored-by: jackwener <jakevingoo@gmail.com>
* feat(xiaoe): add 小鹅通 (Xiaoe-tech) student platform adapter
Add 5 YAML adapters for 小鹅通 (xiaoe-tech.com), the leading Chinese
online education platform:
- courses: list purchased courses with URLs and shop names
- detail: course info (name, price, user count, shop)
- catalog: full course outline supporting normal courses (type 50),
columns (type 6), and big columns (type 8)
- play-url: get M3U8 play URL via direct API for video courses,
and Vue component tree search + Performance API polling for
live replay courses
- content: extract rich-text page content as plain text
Technical notes:
- Strategy: cookie (reuses Chrome login session)
- Framework: Vue 2 + Vuex Store (SPA)
- Video courses use a two-step API chain:
detail_info.get → play_sign → getPlayUrl → M3U8
- Live replays use Performance API + Vue data tree polling
- Catalog expands chapters via Vue component method getSecitonList()
- Supports multiple stores (cross-domain cookie sharing via
study.xiaoe-tech.com)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* review: stop truncating xiaoe content
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-authored-by: jackwener <jakevingoo@gmail.com>
jackwener