- add absolutePath field to CliproxyErrorLog interface
- inject absolute path in routes.ts from getCliproxyWritablePath()
- update copy button to use absolute path with fallback to filename
- refactor error-logs-monitor to remove demo mode
- add error-log-parser lib for structured log parsing
- check cached update result synchronously on every command
- show warnBox notification if update available
- skip check for version/help/update commands
- refresh cache in background if stale (>24h)
- use proper UI components (warnBox) for notification
Key changes:
- Changed default model from claude-sonnet-4.5 to gpt-4.1 (free tier compatible)
- Updated all Claude models to minPlan: 'pro' (requires paid Copilot subscription)
- Replaced 'localhost' with '127.0.0.1' for more reliable local connections
(bypasses DNS resolution and potential IPv6 issues)
- Updated UI presets: replaced Claude Haiku with Raptor Mini in free tier
- Fixed FREE_PRESETS to only include actually free models
This fixes the "model_not_supported" error when using free tier GitHub Copilot
subscription, which doesn't have access to Claude models.
Changed copilot command routing logic to only intercept known subcommands
(auth, status, models, start, stop, enable, disable, help). When user runs
bare 'ccs copilot' without subcommands, it now falls through to profile
detection and executes Claude with the Copilot API backend.
This allows:
- `ccs copilot auth` → command handler (authenticate)
- `ccs copilot status` → command handler (show status)
- `ccs copilot` → profile execution (run Claude with Copilot)
Make the "use at your own risk" warning more transparent:
- UI: Expanded warning banner with clear "Unofficial API" heading
- UI: Added explicit no warranty/no responsibility text
- Config YAML: Updated comments with !! DISCLAIMER !! section
- Types: Updated CopilotConfig JSDoc with full disclaimer
CCS provides no warranty and accepts no responsibility for
consequences of using this reverse-engineered GitHub Copilot API.
- Add ProjectSelectionHandler to parse and handle CLIProxy project prompts
- Integrate project selection into AuthHandler with UI-first support
- Add project selection API endpoints and WebSocket events
- Implement ProjectSelectionDialog and useProjectSelection hook in UI
- Fixes macOS OAuth timeout by handling interactive project selection via UI
Previously, WebSearch status showed "Ready (Gemini)" when the CLI binary
was installed, misleading users who hadn't authenticated yet.
Changes:
- Add isGeminiAuthenticated() to check ~/.gemini/oauth_creds.json
- Add 'needs_auth' state to WebSearchReadiness type
- Show warning "[!] Gemini: run 'gemini' to login" when not authenticated
- Fix incorrect 'gemini auth login' references (no such command exists)
- Update docs with correct npm install and authentication instructions
- add hasTokenFile() for fast auth check without subprocess
- capture device code from copilot-api auth output
- export AuthFlowResult type with deviceCode and verificationUrl
- echo auth output to terminal while capturing for UI
- Add getCliproxyWritablePath() helper function
- Set WRITABLE_PATH env var when spawning CLIProxy in both executors
- Logs will now be stored in ~/.ccs/cliproxy/logs/ instead of CWD
- Enables error log viewer to find logs in predictable location
Note: CLIProxyAPI still has hardcoded MaxBackups=0 (unlimited).
Log rotation should be addressed in CLIProxyAPI upstream.
- POST /api/copilot/install: install copilot-api (optional version)
- GET /api/copilot/info: get install status, version, path
- GET /api/copilot/status: include version in response
- copilot-auth: use getCopilotApiBinPath() instead of npx
- copilot-daemon: move PID file to ~/.ccs/copilot/daemon.pid
- copilot-executor: auto-install copilot-api on first use
- index: export package manager functions
- install copilot-api locally to ~/.ccs/copilot/
- auto-update with version caching (1hr TTL)
- version pinning support via .version-pin file
- mirrors CLIProxy binary-manager pattern
Add ErrorLogsMonitor component to Home page that displays CLIProxyAPI
error logs when requests fail. Users can now diagnose why success rates
drop by viewing detailed error log contents.
Backend:
- Add fetchCliproxyErrorLogs/fetchCliproxyErrorLogContent in stats-fetcher
- Add GET /api/cliproxy/error-logs and /api/cliproxy/error-logs/:name routes
- Include path traversal protection for filename validation
Frontend:
- Add CliproxyErrorLog type and errorLogs API methods
- Add useCliproxyErrorLogs/useCliproxyErrorLogContent hooks
- Create ErrorLogsMonitor component with expandable log viewer
- Integrate into Home page below AuthMonitor
Closes#132
CLIProxy now persists by default after terminal closes:
- Spawn proxy with detached mode (survives terminal close)
- Remove auto-kill on Claude exit (proxy keeps running)
- Add `ccs cliproxy stop` to explicitly terminate proxy
- Add `ccs cliproxy status` to show running proxy info
Closes#129
WebSearch hook was blocking native Claude accounts (ccs ck, default)
with "No Providers Enabled" message instead of passing through to
server-side WebSearch.
Changes:
- Add shouldSkipHook() detection for account/default profile types
- Fix blocking bug: exit(0) when no providers enabled instead of blocking
- Pass CCS_PROFILE_TYPE and CCS_WEBSEARCH_SKIP env vars for profile-aware behavior
- Add profile type signals to settings, cliproxy, account, and default execution paths
- add GET/PUT /api/copilot/settings/raw for copilot.settings.json
- update PUT /api/copilot/config to handle model mapping fields
- sync model mappings between settings file and unified config
- add opus_model, sonnet_model, haiku_model to CopilotConfig
- update generateCopilotEnv to use tier-specific models
- fall back to default model when tier not configured
- GET/PUT /api/copilot/config for configuration
- GET /api/copilot/status for integration status
- POST /api/copilot/auth/start for OAuth flow
- GET /api/copilot/models for available models
- POST /api/copilot/daemon/start|stop for daemon control
- add 'ccs copilot' subcommand with auth, status, models, start, stop, enable, disable
- route 'ccs copilot' as profile name to copilot executor
- integrate copilot command into main CLI entry point
- Removed MAX_THINKING_TOKENS (8191) limit for Claude models via agy.
- Updated default agy haiku model to gemini-3-flash-preview.
- Updated unit tests to match new model expectations.
Add session tracking with reference counting for CLIProxy instances.
Multiple CCS sessions now safely share a single proxy on port 8317.
When any session exits, proxy only terminates if it was the last session.
Changes:
- Add session-tracker.ts: lock file based reference counting
- Modify cliproxy-executor.ts: reuse existing proxy, conditional cleanup
- Add 21 unit tests for session tracker functionality
Fixes#118
- Add model field to Gemini/OpenCode WebSearch providers
- Config.yaml now explicitly stores model (gemini-2.5-flash, opencode/grok-code)
- UI: blur-to-save for model inputs (no save on every keystroke)
- UI: collapsible install hints inside provider cards
- UI: left accent border for enabled providers (better light theme contrast)
- Hook: use systemMessage for info-styled confirmation
- Hook: cleaner result format without defensive "NOT an error" language
- Hook timeout now computed from max provider timeout + 30s buffer
- Removes hardcoded HOOK_TIMEOUT_SECONDS constant
- ensureHookConfig() now updates timeout when it changes
- Default OpenCode model changed from gpt-5-nano to grok-code
- Default OpenCode timeout changed from 60s to 90s
- Single source of truth: config.yaml controls all timeouts
Major refactor of WebSearch hook to:
1. **Respect config.yaml settings**: Hook now reads CCS_WEBSEARCH_GEMINI,
CCS_WEBSEARCH_OPENCODE, CCS_WEBSEARCH_GROK env vars to determine which
providers to use. Only enabled AND installed providers are tried.
2. **Consolidate prompts/models**: Added PROVIDER_CONFIG section at top of
hook file for easy prompt engineering:
- gemini: model + prompt template
- opencode: model (overridable via env) + prompt template
- grok: model + prompt template
3. **Pass provider states via env**: Updated getWebSearchHookEnv() to pass
individual provider enabled states as env vars.
Breaking change: Hook no longer falls back to all installed CLIs. It now
strictly respects config.yaml settings.
The mergeWithDefaults() function was only merging gemini provider config,
completely ignoring opencode and grok providers. This caused:
1. OpenCode enabled=true in config.yaml being reset to false
2. Dashboard showing OpenCode toggled off after reload
3. "Disabled in config" message when only OpenCode was enabled
Fixed by adding opencode and grok to the mergeWithDefaults() function
and updating getWebSearchConfig() to check if ANY provider is enabled.
Previously, ensureHookConfig() would return early if any WebSearch hook
existed in settings.json, even if it pointed to an old/renamed file.
Now it checks if the command path matches the expected hook and updates
it if needed.
This fixes the "No Providers Configured" error when the hook file was
renamed from websearch-gemini-transformer.cjs to websearch-transformer.cjs.
Add automatic fallback chain (Gemini -> OpenCode -> Grok) when multiple
WebSearch CLI providers are available. The hook now tries each provider
in order until one succeeds.
Changes:
- Update websearch-transformer.cjs with tryOpenCodeSearch() and tryGrokSearch()
- Implement fallback chain logic in processHook()
- Add outputAllFailedMessage() for when all providers fail
- Update outputNoToolsMessage() with all three install options
- Update config.yaml comments to document OpenCode and fallback behavior
- Add OpenCodeWebSearchConfig type with enabled, model, timeout options
- Implement getOpenCodeCliStatus() for CLI detection with caching
- Add hasOpenCodeCli() and clearOpenCodeCliCache() helper functions
- Update getWebSearchCliProviders() to include OpenCode provider info
- Update hasAnyWebSearchCli() to check all three providers
- Update getCliInstallHints() with OpenCode install command
- Update WebSearchStatus interface to include opencodeCli
- Update getWebSearchReadiness() to report OpenCode status
- Update routes.ts PUT /api/websearch to handle opencode config
- Update routes.ts GET /api/websearch/status to return opencodeCli
- Add OpenCode provider card in settings.tsx UI
- Add purple-themed installation hints for OpenCode
- Order providers: Gemini (1st) -> OpenCode (2nd) -> Grok (3rd/last)
OpenCode uses `opencode run` with `opencode/gpt-5-nano` model for
web search. FREE tier available via OpenCode Zen, no API key required.
- Update WebSearch section comments in unified-config-loader.ts
- Remove outdated MCP references (web-search-prime, brave, tavily, customMcp)
- Add accurate CLI provider info (Gemini CLI FREE, Grok CLI paid)
- Add GrokWebSearchConfig type to unified-config-types.ts
- Update PUT /api/websearch route to handle both gemini and grok
Old comments listed MCP-based providers that were never implemented.
New comments accurately reflect the CLI-based WebSearch implementation.
- Update install command to npm install -g @vibe-kit/grok-cli
- Update docs URL to github.com/superagent-ai/grok-cli
- Fix env var from XAI_API_KEY to GROK_API_KEY
- Update description to reflect AI coding agent capabilities
- Add Grok CLI detection (grok-4-cli by lalomorales22) alongside Gemini CLI
- Add WebSearch health check group to health-service.ts
- Update settings UI to show both Gemini and Grok CLI providers
- Add detailed install hints when no WebSearch CLI is installed
- Update API routes to return grokCli status
- Both CLI and dashboard users now see clear installation guidance
Providers:
- Gemini CLI: FREE tier (1000 req/day), no API key needed
- Grok CLI: Requires xAI API key (XAI_API_KEY), web + X search
kaitranntt