Claude Code
Connect Claude Code to Passion8. Install the CLI, desktop app, and official IDE extensions on macOS or Windows, then configure Passion8 manually or with CC-Switch.
Claude Code is Anthropic's agentic coding tool for terminals, IDEs, desktop, and web. Connecting it to Passion8 routes its Anthropic-compatible requests to https://passion8.cc.
Claude Code uses https://passion8.cc as its Base URL, without /v1. Codex and OpenAI-compatible clients use https://passion8.cc/v1.
#Connection steps
Install first, then configure. Choose either manual configuration or CC-Switch with explicit Passion8 settings.
InstallationInstall the CLI, desktop app, and official IDE extensions on macOS or Windows.ConfigurationConnect to Passion8 manually or with CC-Switch. The Base URL has no /v1 suffix.
#Learning and workflows
After connecting successfully, learn how Claude Code works, its command system, and team workflows.
How it works and extension mapThe agentic loop, models, tools, context loading, extensions, and cache costs.Official coverage mapAn audit of 163 official Markdown pages, 101 slash commands, and local Chinese documentation entry points.Command referenceCLI flags, slash commands, background agents, reviews, shortcuts, and permission modes.Commands and cache effectsCommand-by-command effects on 5-minute/1-hour TTL, models, effort, fast mode, MCP, and background agents.Interactive mode and terminal experienceKeyboard shortcuts, Vim mode, shell mode, background Bash, tmux, fullscreen, voice input, and caching.CLI and terminal referenceCLI commands, installation channels, environment variables, PATH, terminal shortcuts, fullscreen, voice, and deep links.Best practices and workflowsConnect exploration, planning, implementation, verification, review, subagents, and headless scripts into executable workflows.Prompt library and team adoptionReusable prompts for exploration, implementation, debugging, review, releases, champions, and rollouts.
#Context and sessions
Models, effort, and fast mode/model, /effort, opusplan, fallback, Advisor, 1M context, and fast mode.Feature availability and provider differencesCapability boundaries for Claude subscriptions, Anthropic API, Passion8, and third-party providers.Memory and rulesCLAUDE.md, auto memory, .claude/rules, and AGENTS.md compatibility.Output stylesDefault, Proactive, Explanatory, Learning, custom output-styles, and cache effects.Context window/context, file reads, rules, hooks, compact, and context usage control.Large codebases and monoreposStartup directories, layered CLAUDE.md, Read deny, sparse worktrees, cross-package access, and directory-level skills.Session managementBoundaries of resume, branch, rename, transcripts, /cd, /clear, and /compact.Checkpoints and rewind/rewind, restoring checkpoints, restoring only code or conversation, and boundaries with Git.
#Security and governance
Permissions and modesallow/ask/deny and default/plan/auto/dontAsk/bypassPermissions rules.Auto mode policy referenceTrusted infrastructure, allow/soft_deny/hard_deny, classifyAllShell, and denial reviews.Sandboxing and isolated environmentsBash sandboxing, Sandbox runtime, Dev Containers, Docker, VMs, network, and credential boundaries.Sandbox and Dev ContainersCompare sandboxes, Dev Containers, custom containers, VMs, credential protection, and cache boundaries.Enterprise deployment overviewPlan organization deployment by provider, authentication, managed settings, permissions, monitoring, and data boundaries.Enterprise networking and administrationProxies, CAs, mTLS, server-managed settings, endpoint-managed settings, and managed MCP.Enterprise controls and managed settingsServer-managed settings, endpoint-managed settings, managed MCP, auto mode, analytics, and compliance checklists.Enterprise rolloutRoll out providers, gateways, settings, MCP, observability, auto mode, and training in stages.Adoption and communications kitChampions, launch announcements, drip campaigns, FAQs, pilot feedback, and Passion8 messaging.Data usage and privacyTraining policies, retention, local plaintext transcripts, telemetry, WebFetch, and Passion8 gateway boundaries.Zero Data RetentionZDR coverage, disabled features, local transcripts, gateway logs, and caching misconceptions.SecurityAPI keys, permissions, hooks, MCP, and gateway security essentials.
#Extensions and automation
Tool referenceRead/Edit/Bash/Grep/Agent/Monitor tools, permission rules, and hook matchers.SubagentsBuilt-in Explore/Plan/general-purpose, custom agents, permissions, models, and context isolation.Parallel agents, background work, and workflowsAgent View, /background, /batch, /fork, /goal, /loop, and dynamic workflows.Advanced workflows and cloud collaborationChoosing Advisor, Agent View, /goal, dynamic workflows, Routines, Ultraplan, and Ultrareview.Agent Teams and channelsCollaboration across Claude sessions, external events, permission relay, and channel MCP boundaries.Routines and /loop schedulingLocal loops, Cron, cloud routines, /schedule, API/GitHub triggers, and TTL strategies.Parallel workspaces with worktrees--worktree, PR worktrees, .worktreeinclude, cleanup strategies, and subagent isolation.Connecting MCP toolsHTTP, SSE, stdio, WebSocket, OAuth, scopes, tool search, and gateway considerations.Gateways and protocolsPassion8 Base URL, header/body forwarding, model discovery, cache fields, and common 400/401 errors.Gateway protocol launch checklistProtocol formats, headers/body, attribution, model discovery, spend limits, error semantics, and cache acceptance checks.Gateway operationsClaude apps gateway, common protocols, OIDC, Postgres, spend limits, model discovery, and cache forwarding.Deploying Claude apps gatewayPrivate ingress, OIDC, Kubernetes, Cloud Run, GCP, Postgres, health checks, and upgrades.Hook automationLifecycle automation with PreToolUse, PostToolUse, ConfigChange, FileChanged, and more.Plugins and skillsPlugin manifests, skills, agents, hooks, MCP, LSP, and marketplace distribution.Plugin marketplaces and distributionOfficial, community, and internal marketplaces, version constraints, recommendation rules, and security plugins.Status line/statusline, context usage, cost, Git, PRs, rate limits, and local script performance.Headless automationclaude -p, --bare, JSON/stream-json, CI scripts, structured output, and background waiting.Agent SDKEmbed the Claude Code agent loop into TypeScript/Python products.Agent SDK API referenceTypeScript/Python installation, query, ClaudeSDKClient, tool helpers, session APIs, settings resolution, and migration.Agent SDK agent featuresAgent loop, settingSources, sessions, skills, subagents, TODO tracking, checkpointing, and hosting limits.Agent SDK capability matrixPermission evaluation, hooks, MCP, SessionStore, cost tracking, OTEL, secure deployment, and caching.Agent SDK runtime patternsCustom tools, system prompts, streaming, structured output, tool search, user approval, and slash commands.Agent SDK streaming and pluginsStreaming input, single messages, plugin loading, skill namespaces, permission pauses, and cache effects.Agent SDK production deploymentSession storage, tenant isolation, OTEL, cost tracking, structured output, tool search, and 5m/1h caching.Interactive pages with artifactsPublish Claude Code output as interactive pages and use local alternatives with Passion8.
#Costs and troubleshooting
Monitoring and analyticsOpenTelemetry, analytics dashboards, traceparent, privacy settings, and Passion8 observability boundaries.Prompt caching5-minute and 1-hour TTL, invalidation, MCP, Passion8 gateway behavior, and usage fields.Cost optimizationCost strategies for /clear, /compact, /context, models, MCP, and hooks.Error referenceRuntime API/HTTP, 529, 500, rate limits, authentication, network, oversized requests, model, and thinking errors.TroubleshootingAuthentication, Base URL, models, settings, hooks, MCP, permissions, cache, and safe mode diagnosis.
#Product entry points and boundaries
Desktop and IDE extensionsBoundaries between Claude Desktop, Claude Code Desktop, IDE extensions, and MCP bridges.Desktop, Web, and IDE workflowsDesktop quickstart, Linux beta, scheduled tasks, Web, Remote Control, VS Code, JetBrains, and Slack.Platforms and integrationsEntry-point boundaries for CLI, VS Code, JetBrains, Desktop, Web, Slack, CI/CD, Code Review, Ultraplan, and Ultrareview.Code Review and CIOfficial Code Review, Ultrareview, Ultraplan, GitHub Actions, GitLab CI/CD, and Passion8 connection boundaries.Chrome and computer useBrowser automation, macOS desktop control, permissions, security boundaries, and cache effects.Web, Remote Control, and mobileClaude Code on the web, Remote Control, /teleport, /schedule, and Passion8 boundaries.GlossaryAgentic loop, MCP, hooks, skills, subagents, channels, ZDR, worktrees, and caching terms.SecurityAPI keys, permissions, hooks, MCP, and gateway security essentials.
Need an account or balance? Start with Quickstart and Top-ups and billing.
#Accounts, costs, and troubleshooting
If you do not have an API key, follow Quickstart to create an account and token. See Top-ups and billing for balance and charging details; the client's official subscription and your Passion8 API balance are separate. For authentication, Base URL, or unavailable-model issues, see FAQ. The console is authoritative for currently available models and prices.
Support
Need help?
For setup, billing, or model issues, email us. Check the status page for uptime.
WeChat / QQ support is available at the bottom right.
