Skip to main content

brg_framework

Shared knowledge for the Block Research Group packages:

LLMs (Claude, Codex, Antigravity, Kimi, Mistral):

  • AGENTS.md - entry point
  • skills - for devs
  • server - for users/students
  • theory - text context about solvers (vector db, rag, links to papers, books)
  • evals - evaluation of the LLM interface

Install

pip install brg_framework

Layout of this repository

brg_framework/
├── AGENTS.md                      # for people and agents editing this repository
├── README.md                      # this file
├── pyproject.toml                 # pip install brg_framework
├── CHANGELOG.md
├── LICENSE
├── src/brg_framework/             # the package: version, list of BRG packages, the brg command
│   ├── __init__.py
│   └── cli.py
├── tests/                         # tests of the package
├── .github/workflows/             # build: checks on every push; release: PyPI on a v* tag
├── theory/                        # one folder per method: fd, dr, tna, tno, dem, ags
│   └── fd/
│       ├── index.md               # what, when, assumptions, limits
│       ├── formulation.md         # equations with stable ids
│       ├── glossary.md
│       ├── references.md
│       └── teaching/              # exercises, notebooks, slides
├── skills/
│   └── brg-methods/
│       ├── SKILL.md               # which method, which package, where it is
│       └── scripts/
│           └── copy_example.py    # copies a BRG example as a .py file
├── server/                        # the MCP server (FastMCP)
│   ├── INSTALL.md                 # add the server to each app
│   ├── pyproject.toml
│   ├── src/brg_mcp/
│   │   ├── server.py              # every .md page below is one command
│   │   ├── teaching.py            # brg-publish: notebooks to the teaching website
│   │   ├── fd/                    # force density pages and examples
│   │   ├── tna/                   # thrust network analysis pages and examples
│   │   ├── compas_fd/
│   │   └── compas_tna/
│   └── tests/
└── evals/                         # tasks and results, still empty

The rules

  • Links. This repository does not describe how to use each package. It only says which package to use and where it is.
    • The brg-methods skill picks the package and gives its repository and docs.
    • If the package ships its own skill or AGENTS.md, the agent follows it.
    • If not, the agent reads the package itself: README, docs, examples, tests and the docstrings of the installed version.
    • What a package ships is up to its developers.
  • One source. Knowledge is Markdown files to be available to chatbots.
  • The installed package is the API truth. The agent checks the installed version and reads its docstrings.
  • One MCP server. For learners it returns the full BRG text.
  • Skills follow the Agent Skills specification. The name is lowercase with hyphens and equals the folder name. The description says what it does and when to use it.
  • Every example in this repository is a real .py file that CI runs against a pinned package version.

Users

Learners skills/ Coders server/
Who Students and others who want to read theory, see examples and do exercises. They do not write a project People writing Python with BRG packages
What they ask "Explain force density." "Show me a hypar example." "Give me exercise 2." "Form-find this net with compas_fd." "Why does my solver fail?"
Door The MCP server The brg-methods skill, then the package
Why Nothing to install. Always the newest text. The full answer appears in the chat, not a link Works with the installed package version. The agent reads only what it needs
Extras Links to notebooks on the teaching website A script that copies examples as files

Installation Users - MCP

Claude

  • Click your name (bottom left) → Settings → Connectors. On some versions it's the left menu → Plugins.
  • Click Add → Add custom connector.
  • Name: BRG. URL: https://brg-mcp.fastmcp.app/mcp. Leave the OAuth fields empty, then click Add.
  • Start a new chat and type mcp_brg.
  • You should see the "BRG methods" menu.

Installation Devs - Skills

Claude

npx skills add petrasvestartas/brg_framework -g -y -a claude-code

Codex

npx skills add petrasvestartas/brg_framework -g

MCP server

  • Lives in server/ in this repository, built with FastMCP.
  • Hosted at one web address. Each student adds it to their own account.
  • Reads the same Markdown files. It holds no content of its own.
  • Returns the full text, and the chat shows it as the answer.
  • No per-package servers.

Scripts and hooks

Goal: copy code as it is, without the model retyping it. A script that copies a file costs almost no tokens and no waiting.

Script (works now). skills/brg-methods/scripts/copy_example.py copies a BRG example into the current folder. Only the Python standard library. It installs with the skill, and the skill tells the agent to run it.

python copy_example.py                    # list the examples
python copy_example.py tna_cross_vault    # write tna_cross_vault.py

Hooks (planned, not built). The agent app would run the same script by itself, without a model call. Build them only if the measurements below show they save something.

Measure first

Compare setups on the same tasks: package only, the server, the skill, the skill and the server. Record input, cached and output tokens, model calls, wall time, and pass or fail. Judge correctness first, then tokens, then time. No saving is claimed until these numbers exist.

References

Metadata

Release files for brg-framework 0.1.2

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for brg-framework 0.1.2
File Size Uploaded Provenance Actions
brg_framework-0.1.2.tar.gz 5.9 MB Without provenance Details for brg_framework-0.1.2.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for brg-framework 0.1.2
File Interpreter ABI Platform Actions
brg_framework-0.1.2-py3-none-any.whl Python 3 none any Details for brg_framework-0.1.2-py3-none-any.whl

Total release size: 6.0 MB

Release files / brg_framework-0.1.2.tar.gz

Download URL brg_framework-0.1.2.tar.gz
Size 5.9 MB
Tags Source
SHA-256 checksum
How to use checksums
25899807a8b73ea52be4590f871339ae5a0dac657145dcf0ef202d168ed8bdc9
BLAKE2b-256 checksum
How to use checksums
682d023dfb99edad875a1b9ff9c118805ae39963ae89a471de071bdbe62e2bdb
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via uv/0.13.9 {"installer":{"name":"uv","version":"0.13.9","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Ubuntu","version":"26.04","id":"resolute","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":true}

Release files / brg_framework-0.1.2-py3-none-any.whl

Download URL brg_framework-0.1.2-py3-none-any.whl
Size 9.9 kB
Tags Python 3
SHA-256 checksum
How to use checksums
4ed581ce5627add82e7a9ebc680805915cebdcd579ca6ae5e932a2162353849e
BLAKE2b-256 checksum
How to use checksums
fe4d2f851528790cb281282ab68c092ffec77b303d282a95986dc8e11eac8c6c
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
Yes
Uploaded via uv/0.13.9 {"installer":{"name":"uv","version":"0.13.9","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Ubuntu","version":"26.04","id":"resolute","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":true}

Security for brg-framework 0.1.2

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page