AI Provider Manager

A VS Code extension for managing AI coding tool providers — quota monitoring, provider switching, session analytics, and skill management.

Features

Quota Monitoring

Real-time API quota/balance display in the status bar for multiple AI providers:

• Token Plan — GLM, Kimi, MiniMax (utilization percentage with visual progress bar)
• Balance — DeepSeek, StepFun, SiliconFlow, OpenRouter, Novita (currency amount)
• General / NewAPI / Custom — arbitrary endpoints via configurable templates
• GitHub Copilot — auto-auth via VS Code GitHub Auth (no manual API key)
• Per-provider refresh intervals with health degradation (3x backoff after 3 consecutive failures)
• Usage history recording (24h retention, up to 1440 data points per provider)
• Compact multi-provider display (collapses to count when >3 providers)

Provider Management

• Add, edit, delete, and switch between multiple AI providers
• 6 query templates with auto detection based on base URL
• Config import/export (JSON, API keys excluded for security)
• Local config scanning — auto-discovers existing tool configurations on first startup
• Single-slot matching — detects which provider is currently active in single-slot tools
• In-config indicators — shows which providers are actually written to tool config files
• Claude Code common config — toggle hide attribution, agent teams, tool search, effort level, auto-upgrade

Multi-Tool Support

Model Testing

Built-in streaming model connectivity test supporting Anthropic, OpenAI Chat, OpenAI Responses, and Gemini Native API formats. Reports HTTP status, response time, and model used.

Session Manager

Browse and analyze sessions from 6 tools (Claude Code, Codex, OpenCode, OpenClaw, Gemini CLI, Hermes):

• Session list with message viewer
• Per-session analytics: token usage, cost calculation (per-model pricing), tool stats, model distribution
• Aggregate analytics across sessions with Canvas trend charts (4 series: input, output, cache create, cache read)
• Date range filtering (today, 7d, 30d, all, custom)
• Prompt quality evaluation — 10 rule-based checks (miscommunication, verbosity, repetition, missing context, tool inefficiency, anti-patterns, typos, discontinuity, session scope, token waste)
• LLM-assisted evaluation — optional second-pass quality analysis using your configured AI provider
• Error categorization — 10 error patterns (user interruption, timeout, file not read, etc.)
• Batch delete and multi-format export (JSON, CSV, Markdown, Quality Report HTML)

Skill Manager

Multi-app skill sync management:

• Discover skills from GitHub repos (default 4 repos, customizable) and skills.sh search
• Install/uninstall with multi-app sync (junction symlink preferred, copy fallback, auto mode)
• Supports Claude Code, Codex, Gemini CLI, OpenCode, Hermes
• Local unmanaged skill scanning — finds skills on disk not managed by the extension
• Install from directory or ZIP file (recursive SKILL.md discovery)
• Centralized storage at ~/.ai-provider-manager/ with auto-migration

Omo Integration (Oh-My-OpenCode)

Manage Oh-My-OpenCode/Oh-My-OpenAgent configurations:

• Standard variant (11 agents) and Slim variant (7 agents)
• Enable/disable agents, MCPs, hooks, skills
• Category-based model recommendations

i18n

English and Chinese (zh-cn) localization covering all UI panels and VS Code commands.

Usage

1. Click the AI Provider Manager icon in the sidebar to open the configuration panel
2. Select a target tool → click + Add to add a provider
3. Enter your API Key → click Save
4. Click Switch to activate the provider
5. Status bar shows real-time quota/balance

Commands

Configuration

Build & Test

npm install
npm run compile      # Compile TypeScript
npm run watch        # Watch mode
npm test             # Run unit tests (vitest)

Press F5 in VS Code to launch Extension Development Host for testing.