Create CLI and HTTP adapters for AI models with configuration and testing.
npx skills add https://github.com/blueman82/ai-counsel --skill adapter-factory
Please help me install this Agent Skill. Skill: adapter-factory Source: https://github.com/blueman82/ai-counsel/tree/main/.claude/skills/adapter-factory Command: npx skills add https://github.com/blueman82/ai-counsel --skill adapter-factory
Integrating new AI models or services into the AI Counsel deliberation system requires creating custom adapters, which involves handling subprocesses, HTTP requests, parsing outputs, and managing configuration. This skill provides a comprehensive guide to streamline the process of adding new models.
To create a new CLI adapter, create a file adapters/your_cli.py and define YourCLIAdapter inheriting from BaseCLIAdapter. Implement the parse_output method to extract the model's response from the raw CLI output.
High-intent search queries and answers about installing and using this skill.
Integrating new AI models involves creating custom adapters that handle subprocess execution, HTTP requests, output parsing, and configuration. This Skill provides comprehensive guidance for building CLI and HTTP adapters that allow new models to participate in multi-model debates.
CLI adapters manage command-line tool execution and parse raw output, while HTTP adapters handle API requests, response parsing, and retry logic. Choose CLI adapters for local tools like Ollama and HTTP adapters for remote AI services.
Use environment variable substitution in your adapter configuration to keep API keys and secrets out of config files. This approach isolates credentials from version-controlled code while remaining accessible at runtime.
Yes. Implement unit tests using mocking for CLI adapters and VCR cassettes for HTTP interactions. This lets you verify adapter logic and response parsing reliably without external dependencies.
The Skill abstracts these concerns through base adapter classes. You inherit from BaseCLIAdapter or an HTTP base class and implement core methods like output parsing, reducing boilerplate while handling retries and error cases automatically.
Your new adapter integrates into the deliberation system, allowing the model to receive prompts, execute inference, parse responses, and participate in multi-model reasoning workflows alongside existing integrations.
manutejBuild FastAPI backends with async endpoints, JWT auth, and WebSocket chat.
PthahnixOrchestrate Critic-Defender-Judge debates to validate research artifacts.
PriyanshKuniyalImplement retries, exponential backoff, timeouts, and fault-tolerant decorators in Python.
tiandiyiqiExecute actions across 1000+ services via APIs and SDKs.
