AG2 Files

What's Changed

• fix(mcp): decode resource template URI variables by @asts-top in https://github.com/ag2ai/ag2/pull/3250
• feat(mcp): let a served agent own its HTTP transport settings by @vvlrff in https://github.com/ag2ai/ag2/pull/3252
• feat(tealtiger): add rate_limit governance policy by @nagasatish007 in https://github.com/ag2ai/ag2/pull/3246
• feat(skills): render a MemorySkill body per read by @Lancetnik in https://github.com/ag2ai/ag2/pull/3254
• fix(a2a): require only the chosen transport's SDK extra by @vvlrff in https://github.com/ag2ai/ag2/pull/3257
• fix(subagents): lock max_concurrency semaphore provisioning per loop by @AmirF194 in https://github.com/ag2ai/ag2/pull/3243
• fix(mcp): follow pagination when discovering tools by @asts-top in https://github.com/ag2ai/ag2/pull/3263
• fix(anthropic): report thinking tokens, and keep a measured zero a zero by @vvlrff in https://github.com/ag2ai/ag2/pull/3259
• feat(events): retain the provider's response identifier by @vvlrff in https://github.com/ag2ai/ag2/pull/3260
• feat(openai): expose prompt cache controls for chat and responses by @vvlrff in https://github.com/ag2ai/ag2/pull/3261
• feat(web-fetch): gate Anthropic-only options by version by @vvlrff in https://github.com/ag2ai/ag2/pull/3264
• feat(web-fetch): control which sources supply fetchable URLs by @vvlrff in https://github.com/ag2ai/ag2/pull/3265
• feat (OpenAI): prompt cache diagnostics by @vvlrff in https://github.com/ag2ai/ag2/pull/3266
• refactor(mappers): build builtin tool payloads from the SDKs' param types by @vvlrff in https://github.com/ag2ai/ag2/pull/3270

New Contributors

• @asts-top made their first contribution in https://github.com/ag2ai/ag2/pull/3250
• @AmirF194 made their first contribution in https://github.com/ag2ai/ag2/pull/3243

Full Changelog: https://github.com/ag2ai/ag2/compare/v1.0.5...v1.0.6