v0.2.0

0.2.0 (2026-07-10)

Bug Fixes

• avoid inventing unknown token limits (8d7b5ba)
• construct discovery URLs safely (23daa38)
• honor empty model discovery responses (cbb99ed)
• initialize discovered OpenAI models safely (404cb31)
• map output limits independently (c476ba8)
• merge concurrent cache updates (478b578)
• preserve existing OpenAI model definitions (36c2ae0)
• prioritize provider exclusions (7cfff80)
• publish provenance from a GitHub-hosted runner (5a55613)
• redact discovery URLs in logs (a31e573)
• restrict discovery to HTTP endpoints (04e3685)
• retry discovery after transient failures (729c732)
• scope model cache by credentials (c6f31e1)
• use discovered OpenAI API model IDs (369c34d)
• validate cached model data (3cd8191)
• validate discovery response models (a0e2ab1)
• write model cache atomically (ede8b60)

Features

• export discovery option types (28bc77b)
• honor platform cache directories (3c24937)
• limit discovery response size (d4e3902)
• log model discovery failures (acc461c)
• make discovery timeouts configurable (427e40d)
• map common model metadata aliases (34b0394)
• support bounded model pagination (a81a67c)
• support custom discovery headers (d41d8a5)
• validate discovery options (cb41841)