v0.3.0

This release adds Starknet ecosystem knowledge support, Grok web/X search for real-time context, conversation suggestions, reasoning streaming, source mapping with citations, Gemini embeddings, website documentation crawler, and ingester stability improvements.

What's Changed

• feat: store chunk sourceMap when available by @enitrat in #63
• feat: emit sources in streaming mode by @enitrat in #64
• feat: add starknet-agent generation mode by @enitrat in #66
• feat: add suggestion entrypoint by @enitrat in #67
• dev: use gemini embeddings by @enitrat in #68
• refactor: use dspy.context embedder by @enitrat in #69
• feat: support tracing on streaming endpoints by @enitrat in #70
• feat(summarizer): support from summarizing from crawled website by @enitrat in #71
• fix(ingester): handle malformed input, support source mapping by @enitrat in #72
• feat(ingester): support metadata updates only by @enitrat in #73
• dev: update doc rater system prompt by @enitrat in #75
• feat: send full prediction on streaming final_response by @enitrat in #77
• fix: remove reliance on .env in docker image by @enitrat in #79
• fix: stalled ingesters in non-interactive mode by @enitrat in #81
• dev: remove unused callback by @enitrat in #82
• feat: support reasoning streaming by @enitrat in #83
• dev: reorg codebase structure by @enitrat in #85
• feat: dataset analysis scripts by @enitrat in #84
• feat: add grok search for general knowledge by @enitrat in #86

Docker Images

As with all our previous releases, you can find the Docker images:

• Backend
• Ingester

Full Changelog: v0.2.2...v0.3.0

v0.2.2

This release features architectural improvements with lightweight agent registry, enhanced documentation crawling (OpenZeppelin, Starknet.js), DSPy 3.0 upgrade, performance optimizations, and stability fixes for a more robust RAG system.

What's Changed

• refactor: replace config-based agents with lightweight registry by @enitrat in #37
• feat: add documentation snapshot crawler tool by @enitrat in #41
• feat: add documentation crawler for OpenZeppelin docs by @enitrat in #42
• fix: race condition on VectorDb singleton aquisition by @enitrat in #44
• feat: upgrade to DSPy 3.0 by @enitrat in #45
• refactor: migrate scripts to src/scripts and switch to uv build backend by @enitrat in #47
• dev: update StakrnetDocs ingester to be compatible with new website by @enitrat in #46
• security: add override to prevent malicious error-ex versions by @ijusttookadnatest in #49
• dev: remove custom tracing callback by @enitrat in #50
• dev: disable LLM cache by @enitrat in #51
• fix: streaming mode by @enitrat in #52
• dev: add scripts for dataset extraction by @enitrat in #58
• dev: fix logging config by @enitrat in #57
• dev: speedup starklings compilation in eval script by @enitrat in #55
• feat: GEPA optimizers by @enitrat in #56
• dev: update cairo book source by @enitrat in #54
• dev: remove deprecated backend by @enitrat in #53
• feat: add starknet_js as documentation source by @enitrat in #59
• dev: add g++ install to dockerfile (required for uv build) by @enitrat in #60
• dev: update gemini models by @enitrat in #61

Docker Images

As with all our previous releases, you can find the Docker images:

• Backend
• Ingester

Full Changelog: v0.2.1...v0.2.2

v0.2.1

This release adds LLM-based document retrieval and migrates to environment-only configuration.

What's Changed

• feat: llm judge for docs retrieval by @enitrat in #32
• refactor: migrate to environment-only configuration system by @enitrat in #38
• fix: release workflow by @ijusttookadnatest in #40

Docker Images

As with all our previous releases, you can find the Docker images:

• Backend
• Ingester

Full Changelog: v0.2.0...v0.2.1

v0.2.0 - DSPY Migration

A migration to a Python backend using DSPY.

What's Changed

• fix: oz ingester by @enitrat in #25
• fix: small fix for gemini model and headers by @ijusttookadnatest in #26
• feat: add starklings evaluation report by @ijusttookadnatest in #27
• feat: improve starkling evaluation with feedback by @ijusttookadnatest in #30
• feat: rewrite codebase in Python + DSPY by @enitrat in #29
• fix(refacto): token tracking in response and readme by @ijusttookadnatest in #34
• dev: add pytest in CI by @enitrat in #36
• refactor(tests): setup proper test architecture by @enitrat in #35
• feat(ingester): better markdown splitter by @enitrat in #33

Full Changelog: v0.1.1...v0.2.0

v0.1.1

Cairo Coder v0.1.1

This release focuses on stability improvements and better API structure.

What's Fixed

• Embeddings ingestion for large documentation sources
• Token calculation accuracy for better token usage tracking
• Agent-specific API routes for cleaner architecture

What's Changed

• fix: big embeddings ingestion by @enitrat in #21
• refactor: add agent-specific routes by @enitrat in #23
• fix(agents): get exact token calculation by @ijusttookadnatest in #22

Docker Images

As with all our previous releases, you can find the Docker images:

• Backend
• Ingester

Full Changelog: v0.1.0...v0.1.1

v0.1.0

Cairo Coder v0.1.0 - Initial Release

Cairo Coder is now officially available! This first release brings you the most powerful open-source Cairo language code generation service, built with advanced Retrieval-Augmented Generation (RAG) technology.

Documentation Sources

Cairo Book • Cairo by Example • Starknet Foundry • OpenZeppelin • Core Library • Scarb

What's New

• RAG-powered code generation with comprehensive Cairo documentation
• OpenAI-compatible API for easy integration
• Multiple LLM support (OpenAI, Anthropic, Google)
• Docker deployment ready out of the box
• MCP support for advanced workflows

What's Changed

• core: Cairo agent migration from starknet-agent repo by @ijusttookadnatest in #1
• core(db) : Migration from mongodb to postgresql by @ijusttookadnatest in #2
• feat: support and recommend using gemini 2.5 flash by @enitrat in #4
• feat: add langsmith support by @ijusttookadnatest in #6
• update cairobook ingester to use llms-full file by @enitrat in #7
• feat: add cairo corelib source by @enitrat in #8
• feat(backend): add token tracking by @ijusttookadnatest in #9
• feat(mcp): adding bypass of summarizer for mcp by @antiyro in #11
• clean(docker): clean setup files and doc by @ijusttookadnatest in #13
• fix(docs) added headers to readme by @antiyro in #14
• feat(ci): update and fix ci by @ijusttookadnatest in #12
• fix(trunk): cleanup cache files by @ijusttookadnatest in #15
• feat: add scarb docs by @enitrat in #16
• ci: add release workflow and update dependances by @ijusttookadnatest in #17

New Contributors

• @ijusttookadnatest made their first contribution in #1
• @enitrat made their first contribution in #4
• @antiyro made their first contribution in #11

Docker Images

As with all our releases, you can find the Docker images:

• Backend
• Ingester

Full Changelog: v0.0.0...v0.1.0