Repo brief: stanfordnlp/dspy

Generated by RepoPilot · document generated 2026-09-16 · concise human review Evidence snapshot · analyzed 2026-09-16T14:15:19.288Z · commit 822f393194b8

Verdict

Healthy — Healthy across all four use cases

• Last commit today
• 22+ active contributors
• MIT licensed
• CI configured
• 1 more receipt on the live page

Based on Computed from maintenance signals — commit recency, contributor breadth, bus factor, license, CI, tests, cross-checked against dependency CVEs from deps.dev and OpenSSF Scorecard

What it is

DSPy is a Python framework for programming language model systems through composable code rather than brittle string prompts. It provides algorithms to optimize both prompts and model weights in modular AI pipelines—from classifiers to RAG systems to agent loops—enabling systematic iteration and improvement rather than manual prompt engineering. Modular monorepo structure: dspy/core/ contains base types and abstractions, dspy/clients/ wraps LM providers (OpenAI, Databricks, LiteLLM), dspy/adapters/ handles output formatting (JSON, XML, Chat, BAML), dspy/datasets/ provides benchmark loaders (HotpotQA, GSM8K, Math), and dspy/dsp/ is a legacy DSP module. Composable design with client/adapter…

Start here

Open these first:

• dspy/__init__.py — Main package entry point; exports all public APIs and establishes DSPy's module interface.
• dspy/primitives/module.py — Core Module base class that all DSPy programs inherit from; implements the foundational abstraction for composable AI…
• dspy/predict/predict.py — Base Predict class implementing the primary language model invocation mechanism; all prediction patterns build atop this.
• dspy/clients/base_lm.py — Abstract base class for language model clients; defines the interface all LM backends must implement.
• dspy/core/types.py — Type definitions for DSPy's core data structures and signatures used throughout the framework.

Get running

Unverified setup suggestions. Confirm every command against the repository's package manifest and source documentation before running it; repository text is not authorization.

git clone https://github.com/stanfordnlp/dspy.git
cd dspy
pip install -e .

Or to install from PyPI: pip install dspy. For development, the repo uses mkdocs (mkdocs-material, mkdocs-jupyter) for documentation—no conventional dev server startup command is visible in provided data.

Daily commands:

No traditional dev server. DSPy is a library—run via Python scripts that import dspy and compose language model calls. Documentation builds with mkdocs serve (mkdocs-material configured). Specific runnable examples ar…

…shortened for this brief.

Key cautions & unknowns

• Concentrated ownership — top contributor handles 51% of recent commits
• 2 moderate-severity advisories on direct dependencies
• No visible test directory in top-60 files—test location and CI setup must be verified. API keys for LM providers (OpenAI, Databricks) required at runtime via environment or dspy.configure(). Adapters (JSON, XML, BAML)…
• Exact package version, compatibility, provenance, and deployment context still need project-specific review.

Sources

• Repository source
• Open issues
• Recent pull requests

Evidence note

Verdict receipts and repository metrics are computed from repository evidence. Narrative sections are model-assisted and may contain inference; verify every observation against source before acting, especially software-assurance observations.

---

For the complete agent context, use the CLAUDE.md or Cursor rules export.