Overview
This repository holds connectors for 22 sources of financial and economic data: central banks, statistical offices, and market data vendors. Each source is its own pip package. Install the ones you need and parsimony picks them up automatically. The parsimony agent skill teaches coding agents to work with them.
Quickstart¶
from parsimony_fred import CONNECTORS
fred = CONNECTORS.bind(api_key="...")
print(fred.describe())
result = fred["fred_fetch"](series_id="UNRATE")
print(result.frame.tail()) # tabular results expose .frame
Instead of bind, you can set the FRED_API_KEY environment variable. More in credentials.
Search comes first¶
Getting data is two steps: search for the series you want, then fetch it by the ID the search returned. Every package ships both kinds of connectors. Where the provider has a search API, the search connector calls it directly. Where the provider can only list its contents, we build a search catalog with parsimony's catalog tooling and publish it on Hugging Face; the catalog downloads and caches locally the first time you search.
Available connectors¶
The table is generated from the package metadata. Run make readme-roster to refresh it.
Adding a connector¶
A connector is a small package under packages/. Start with the authoring guide and CONTRIBUTING.md. GOVERNANCE.md covers what we accept and how packages are maintained.
Documentation¶
The complete documentation is published at docs.parsimony.dev.
Related projects¶
- parsimony is the framework these connectors are built on.
- parsimony-agents is a ready-made data analysis agent built on these connectors.
Development¶
make sync # install everything with uv
make verify PKG=fred # lint + types + tests for one package
make verify-all # the same for every package
make verify runs the same checks as CI. See CONTRIBUTING.md.
License¶
Apache-2.0. See LICENSE.