Migrating to 0.2.0

Version 0.2.0 moves token usage parsing from praxis-ai-apis into the private token usage subsystem in praxis-ai-filters.

Proxy configuration

No configuration changes are required. The token_count and token_usage_headers filter names, provider values, and metadata keys remain unchanged.

Rust API

The following 0.1.x API is removed:

praxis_ai_apis::token_usage::{
    TokenUsage,
    TokenUsageProvider,
    set_token_usage,
}

Provider response parsing now belongs to TokenCountFilter and is not exposed as a standalone library API. Add the filter to a pipeline and select the provider through its YAML configuration. Extracted counts remain available through these HttpFilterContext metadata keys:

• token.input
• token.output
• token.total

Custom filters that produce token counts can write the same keys with HttpFilterContext::set_metadata. Code that requires standalone provider parsing must remain on 0.1.x or own that parsing until a new public abstraction is introduced.