Single-mode and page-state-eval scenes now get auto-wrapped into a
default mode and compiled through compile_multi_mode_request. This
eliminates the old browser_script_with_business_logic code path and
ensures all scenes get responsePath extraction, requestTemplate, and
contentType support.
🤖 Generated with [Qoder][https://qoder.com]
Issue #1 (Critical): Replaced eval() in buildModeRequest with a safe
resolveTemplateValue function that only supports args.fieldName and
args['fieldName'] patterns, eliminating XSS/injection vulnerability.
Issue #2 (Critical): Added defensive guard in detectMode that throws a
clear error when MODES array is empty, preventing undefined property access.
🤖 Generated with [Qoder][https://qoder.com]
The NormalizeRulesJson struct was serializing required_fields and filter_null
as snake_case, but the JavaScript template accesses them as requiredFields
and filterNull. Add #[serde(rename = "requiredFields")] and #[serde(rename = "filterNull")]
to fix the systematic snake_case vs camelCase mismatch in multi-mode templates.
🤖 Generated with [Qoder][https://qoder.com]
Add browser_script_with_modes function that generates JavaScript with:
- detectMode() to select mode based on args
- buildModeRequest() with content-type handling (JSON/form-urlencoded)
- normalizeModeRows() with validation rules
- queryModeData() with jQuery + fetch dual HTTP client
Modify browser_script() to check for modes first before falling back
to business_logic or skeleton templates.
🤖 Generated with [Qoder][https://qoder.com]
Add ModeConditionJson, NormalizeRulesJson, ApiEndpointEnhancedJson, and
ModeConfigJson structs to support multi-mode business logic in generated
scripts. Enhance SceneInfoJson with modes, default_mode, and mode_switch_field.
🤖 Generated with [Qoder][https://qoder.com]
- Add page context validation step with deps.validatePageContext
- Change validation failure status from 'error' to 'blocked'
- Add row normalization partial detection
- Use buildArtifact for all return paths consistently
- Add COLUMNS and REPORT_NAME constants for buildArtifact
- Export determineArtifactStatus, COLUMNS, REPORT_NAME in module.exports
🤖 Generated with [Qoder][https://qoder.com]
- Change function signature to accept destructured object parameter
- Integrate determineArtifactStatus for automatic status calculation
- Use REPORT_NAME and COLUMNS constants (to be defined in Task 6)
- Add proper filtering for partial_reasons (exclude api_/validation_ prefixes)
- Add deduplication for reasons array
- Update call site in buildBrowserEntrypointResult to use new signature
🤖 Generated with [Qoder][https://qoder.com]
Add a helper function to determine artifact status based on business
logic conditions:
- blocked: blockedReason exists (validation failed, wrong page context)
- error: fatalError exists (API call failed)
- partial: reasons array non-empty (some data issues)
- empty: no rows returned
- ok: rows exist with no issues
This is Task 3 of 8 in the progressive template enhancement.
🤖 Generated with [Qoder][https://qoder.com]
Enhance defaultDeps with:
- validatePageContext: Checks hostname against expected_domain for page context validation
- queryData: Dual HTTP client with jQuery $.ajax as primary and fetch as fallback
- jQuery preferred for internal pages that typically have it
- Clear error message if neither client is available
🤖 Generated with [Qoder][https://qoder.com]
The previous implementation used `new URL(endpoint.url, window.location.origin)`
which incorrectly constructed URLs based on the current page's origin. This broke
when endpoint URLs were already complete URLs (e.g., http://20.76.57.61:18080/...).
The fix uses endpoint.url directly since LLM extraction already provides complete
URLs. Also changed from GET with query params to POST with JSON body to match
typical API expectations.
🤖 Generated with [Qoder][https://qoder.com]
- Add SceneInfoJson struct with serde rename for camelCase compatibility
- Add ApiEndpointJson and BusinessLogicJson support structs
- Add scene_info_json field to GenerateSceneRequest (backward compatible)
- Rename browser_script to browser_script_skeleton
- Add new browser_script that delegates based on scene_info presence
- Add browser_script_with_business_logic for enhanced script generation
- Update CLI to accept --scene-info-json parameter
- Update usage string to document new parameter
Generated with [Qoder][https://qoder.com]
- Add /select-folder and /select-file APIs using PowerShell dialogs
- Add --target-url parameter to CLI for explicit target URL override
- Redesign Web UI with folder browse buttons for all path inputs
- Add target_url optional input field for specifying target page URL
- Auto-fill scene-id from selected folder name
🤖 Generated with [Qoder][https://qoder.com]
When HTML has no sgclaw-expected-domain meta tag, analyzer now scans
for external script URLs (http:// or https://) and extracts the
domain (host:port) as expected_domain.
Example:
<script src="http://25.215.213.128:18080/a_js/YPTAPI.js"></script>
→ expected_domain = "25.215.213.128:18080"
This reduces manual editing required for third-party scenes.
🤖 Generated with [Qoder][https://qoder.com]
Remove the validation that required target_url, expected_domain, and
entry_script meta tags for ReportCollection scenes. Third-party scenes
do not have these meta tags, so the generator now uses empty values
that users can manually edit in scene.toml.
🤖 Generated with [Qoder][https://qoder.com]
- Add scene_kind field to GenerateSceneRequest
- Update generate_scene_package to use analyze_scene_source_with_hint
- Implement scene_toml_monitoring for simplified monitoring scene manifests
- Add scene_toml router to dispatch to different templates based on SceneKind
- Add --scene-kind CLI option to sg_scene_generate binary
- Add generator_emits_monitoring_template test
- Create monitoring fixture with index.html (no meta tag, tests hint param)
Monitoring templates have simplified scene.toml without org/period resolvers.
🤖 Generated with [Qoder][https://qoder.com]
Remove duplicate implementation in analyze_scene_source() and replace
with simple delegation to analyze_scene_source_with_hint(source_dir, None).
This eliminates ~50 lines of duplicated logic that could drift apart
from the main implementation. Updated the test to verify the new
behavior where sources without scene-kind meta tag default to
ReportCollection instead of being rejected.
🤖 Generated with [Qoder][https://qoder.com]
- Add SceneKind::Monitoring enum variant with from_str/as_str helpers
- Add analyze_scene_source_with_hint function accepting optional scene kind hint
- User hint takes priority over meta tag, defaults to ReportCollection
- ReportCollection requires target_url, expected_domain, entry_script
- Monitoring type has optional fields
- Add test cases for hint parameter behavior
- Update non_report fixture with required meta tags for ReportCollection
🤖 Generated with [Qoder][https://qoder.com]
Use page context, deterministic plans, and direct-skill metadata as the service-owned bootstrap target precedence so callback-host startup no longer relies on line-loss text matching or the old request-url helper.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
- Auto-connect WebSocket on page load in service console
- Settings modal for editing sgclaw_config.json (API key, base URL, model, skills dir, etc.)
- UpdateConfig/ConfigUpdated protocol messages for remote config save
- save_to_path() for SgClawSettings serialization
- ConfigUpdated handler in sg_claw_client binary
- Protocol serialization tests for new message types
- HTML test assertions for auto-connect and settings UI
- Additional pending changes: deterministic submit, org units, lineloss xlsx export, browser script tool, and docs
🤖 Generated with [Qoder][https://qoder.com]
Sends sgHideBrowerserClosePage (best-effort) before sgHideBrowerserOpenPage
on the same bootstrap WebSocket connection. This prevents duplicate helper
pages across process restarts. Also enables hidden domain mode so the helper
page is not visible to users.
🤖 Generated with [Qoder][https://qoder.com]
The close_helper_page function opened a second browser WebSocket
connection during Drop and sent a close command directly via the WS
bypassing the HTTP polling system. This interfered with the browser's
normal state and caused the helper page to fail to open.
The cached_host lift (previous commit) already solves the duplicate
helper page issue within a single process lifetime. The Drop-based
close logic is deferred until a proper cleanup mechanism is designed.
🤖 Generated with [Qoder][https://qoder.com]
The `expected_domain` was removed from args for normalization but never
re-inserted, causing JS scripts to receive empty expected_domain and
report "missing_expected_domain" errors.
🤖 Generated with [Qoder][https://qoder.com]
Bring the main-branch lineloss default-period fix into feature/claw-ws while keeping the ws submit/backend path intact.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Default month/week deterministic lineloss requests to the source page's built-in time ranges while preserving explicit-period parsing and existing routing contracts.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Keep ws-backed submit flows routing Zhihu natural-language requests through orchestration before direct submit so sg_claw service console behavior stays consistent.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Keep Zhihu hotlist export requests on the orchestration path so natural-language submits without page context no longer fail in direct-submit routing.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Keep the ws submit path while bringing over main's deterministic lineloss routing and the focused merge verification updates.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Add the deterministic tq-lineloss routing and normalization flow so exact-suffix requests execute through the existing browser-script seam with canonical org and period arguments.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Keep browser-attached workflows on the configured direct-skill path and align the Zhihu export/browser regression contracts with the current ws merge state.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Keep the ws branch focused on websocket and Zhihu behavior by dropping staged scene-routing artifacts and restoring single-path skills dir semantics.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Treat direct skill report-artifact payloads as task outcomes so partial and empty reports stay successful while blocked and error statuses fail explicitly.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Add fixed direct-submit skill loading from configured staged skills and validate directSubmitSkill early so malformed configs fail before routing.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Stabilize the service console and callback-host websocket paths so idle disconnects and mid-task client drops no longer wedge task execution or spam repeated commands.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Add registry-driven scene routing and multi-root skill loading so fault-details and 95598 scene skills can be triggered from natural language while still running through the browser-backed runtime.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Consolidate the browser task runtime around the callback path, add safer artifact opening for Zhihu exports, and cover the new service/browser flows with focused tests and supporting docs.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Keep Zhihu browser-attached execution on the callback-host path so direct routes, runtime wiring, and service startup stay aligned for the current websocket browser flow.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Wire the service/browser runtime onto the websocket-driven execution path and add the new browser/service modules needed for the submit flow and runtime integration.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
