Open-source Rust engine that filters and compresses the context AI coding tools read, with MCP server, shell hooks and local audit records.
3.9K356RustApache-2.0Updated 14 hours ago
teaql/teaql-agent-kit
A model-driven harness that gives AI coding agents a model-first workflow: model the business, evaluate and repair it, generate a typed application, and verify the result.
- Stars
- 2,818
- Forks
- 956
- Open issues
- 1
- Last commit
- 16 hours ago
What is it?
TeaQL Agent Kit is a set of agent instructions and tooling for building business applications with the TeaQL framework. Instead of having a coding agent go directly from a requirement to code, it first produces an inspectable domain model in KSML, runs a deterministic evaluation service that reports errors, warnings and repair guidance, and only then generates a typed domain library and runnable workspace for the chosen language.
The generated library exposes typed Q query and E expression APIs, and the agent can request progressive, model-aware Assist for the exact operations it needs rather than reading the whole library. The project supports Java, Rust, Go, Swift, Python, C#/.NET and TypeScript targets, and includes runtime safeguards such as query purpose and comment fields and audit context on writes.
The repository also contains agent-context-kit, a package for managing context lifetimes through host adapters, including consume-once messages, named instruction blocks and file-read exclusion for generated directories. Adapter coverage varies by coding agent, and the project states that these controls require a supporting adapter.
Key features
- 01Turns requirements into a saved KSML domain model before any application code is written.
- 02Runs deterministic model evaluation that reports errors, warnings and repair guidance.
- 03Generates typed Q query and E expression APIs for the selected language.
- 04Provides progressive, model-aware Assist for entity, action and field-level API guidance.
- 05Includes context lifetime controls and read exclusion for generated directories via agent-context-kit.
Use cases
- Having a coding agent build a small CRUD business application from a plain-language requirement.
- Evaluating and repairing a domain model before generating code for a multi-entity system.
- Generating a typed data-access library for Java, Rust, Go, Python or other supported languages.
- Experimenting with context management and guardrails for AI coding agents.
In collections
Topics
FAQ
- What is teaql-agent-kit?
- A model-driven harness that gives AI coding agents a model-first workflow: model the business, evaluate and repair it, generate a typed application, and verify the result. TeaQL Agent Kit is a set of agent instructions and tooling for building business applications with the TeaQL framework. Instead of having a coding agent go directly from a requirement to code, it first produces an inspectable domain model in KSML, runs a deterministic evaluation service that reports errors, warnings and repair guidance, and only then generates a typed domain library and runnable workspace for the chosen language.
- Is teaql-agent-kit free to use? What license does it have?
- teaql-agent-kit is open source under the MIT license. Check the license terms before using it in commercial products.
- Is teaql-agent-kit still maintained?
- Actively maintained. The last commit to teaql-agent-kit was on Oct 24, 2026.
- What can I use teaql-agent-kit for?
- Having a coding agent build a small CRUD business application from a plain-language requirement. Evaluating and repairing a domain model before generating code for a multi-entity system. Generating a typed data-access library for Java, Rust, Go, Python or other supported languages. Experimenting with context management and guardrails for AI coding agents.
- Where can I get teaql-agent-kit?
- The source code and installation instructions are on GitHub: https://github.com/teaql/teaql-agent-kit
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
Guidelines that teach AI coding agents to write modern Go using current language features and standard library additions from Go 1.0 through 1.27.
3.9K125GoApache-2.0Updated 10 days ago
CLI proxy that filters and compresses shell command output before it reaches an AI coding agent's context, cutting token use.
84.5K+157 this week5.4KRustApache-2.0Updated 2 days ago
Open source AI coding agent that runs in VS Code, JetBrains IDEs, or the terminal, with access to 500+ models.
27.7K+24 this week3.3KTypeScriptMITUpdated yesterday
A coding agent built on pi that routes routine decisions, like permission checks and context pruning, to a small judge model so the main model can focus on code.
54463TypeScriptMITUpdated 14 hours ago
A local-first, evidence-first AI coding terminal that connects LLMs to real project files, tools, permissions, verification and context.
45920TypeScriptApache-2.0Updated 2 months ago