v0.2.21

Changes

• allow mcp versions >=1.7.0, <1.8.0 @mpereira (#138)
• prompt-server sse port configuration @evalstate (#141)

New Contributors

• @mpereira made their first contribution in #138

Full Changelog: v0.2.20...v0.2.21

v0.2.20

What's Changed

• LiteLLM+LangFuse Support: Enhancing how request_params are overridden by @theobjectivedad in #129
• Add more checks for when args parsing is disabled by @kahkeng in #133
• Added MCP server cwd setting by @jpeirson in #131
• Fix/misc by @evalstate in #137...
• otel instance id set
• allow hyphens in tool/server names, namespace refactor (llm) #135
• windows sse server startup fix #121
• increase max_iterations for tool call/llm loop
• add cwd functionality test

New Contributors

• @theobjectivedad made their first contribution in #129
• @kahkeng made their first contribution in #133
• @jpeirson made their first contribution in #131
  Thanks to
• @lizzy-0323 for raising hyphen in server defect
• @wowok-ai for raising windows sse server issue

Full Changelog: v0.2.19...v0.2.20

v0.2.19

OpenTelemetry Support

This release includes opentelemetry support for observing endpoints and MCP tool calls. Check the examples/otel directory for a docker compose file (thanks @SecretiveShell ) for starting Jaeger and setting up telemetry.

Changes

• Feat/otel @evalstate (#123)
• Prevent FastAgent CLI argument parsing conflict when embedded @FruitSwimer (#118)
• Fix syntax errors in sample code by @epli2 in #116
• Prevent FastAgent CLI argument parsing conflict when embedded by @FruitSwimer in #118
• Fix: Handle missing args in MCP server configuration @LouD82 (#113)

New Contributors

• @epli2 made their first contribution in #116
• @FruitSwimer made their first contribution in #118

Full Changelog: v0.2.18...v0.2.19

v0.2.18

What's Changed

• Fix ValueError when parsing empty arguments in tool calls (OpenRouter) by @LouD82 in #102
• Command line option to execute a prompt file
• new fast-agent go option to start an interactive session with selected MCP servers
• fix issue #101 (thanks @kikiya)
• fix issue #98 (thanks @kangbaoxing)
• improve tui handling of /prompt/

New Contributors

• @LouD82 made their first contribution in #102

Thanks to @jlyst for testing and diagnosing issues
Thanks to @SecretiveShell for code reviews, and promptmesh A2A type libraries.

Full Changelog: v0.2.17...v0.2.18

v0.2.17

What's Changed

• Fixes to Structured Response handling @evalstate in #94
• Fix to Orchestrator plan steps. API Change: decorator parameter is now plan_iterations rather than max_iterations to disambiguate.
• Improvements in internal content handling
• Introduction of A2A types for (Card and Skill) for workflow usage.
• Google Gemini (OAI) endpoint support by @evalstate in #95

Full Changelog: v0.2.15...v0.2.17

v0.2.15

Changes

• Fix for "Generic" provider to use correct API keys.

v0.2.14

What's Changed

• Added setup check tool. Run it with fast-agent check.
• Feat/setup checks by @evalstate in #84

Full Changelog: v0.2.12...v0.2.14

v0.2.12

What's Changed

• include gpt-4.1 model aliases by @evalstate in #73

Full Changelog: v0.2.11...v0.2.12

v0.2.11

What's Changed

• Fix base url reads from ENV for Generic and OpenRouter by @Amadeus-Winarto in #59
• enhancement: allow sse_read_timeout to be configurable by @satyamsi in #57
• Update README.md by @evalstate in #60
• add a2a-types by @SecretiveShell in #68
• Feat/quickstart by @evalstate in #69

New Contributors

• @Amadeus-Winarto made their first contribution in #59
• @satyamsi made their first contribution in #57
• @SecretiveShell made their first contribution in #68
• @amroalkhatib for workflow testing

Full Changelog: v0.2.8...v0.2.11

v0.2.8

Changes

• Fix/stderr and prompt ui handling @evalstate (#53)
• openrouter support + docs thanks - @gitanon112
• deepseek support (closing pr) thanks @lightionight

What's Changed

• Fix/stderr and prompt ui handling by @evalstate in #53

Full Changelog: v0.2.7...v0.2.8