anthropic

Anthropic’s release notes for Claude Fable 5.1, covering new features, breaking changes, pricing, model availability, and migration details for API users.

A small Rust-based LLM proxy for forwarding chat, messages, and responses requests to providers with gateway-key handling and optional protocol translation between Anthropic and OpenAI styles.

OpenSmith is a local-first, open-source alternative to LangSmith for tracing Python LLM pipelines. It supports decorators, context managers, console output, autopatching, CLI commands, local SQLite storage, and optional Postgres and OpenTelemetry integrations.

Python library for managing LLM conversation history with cache-friendly message ordering, dynamic context, and compaction.

FastAPI backend for the Brush Up tutor app. It retrieves from a markdown note corpus, expands context through graph neighbors, and calls Anthropic to generate grounded programming help for Python, Ruby, and JavaScript.

Minimal Erlang library for calling OpenAI and Anthropic models, with tool-use support and a built-in JSON encoder/decoder.

A GitHub repository of system-prompt overrides for Claude Code, aimed at shrinking the always-on prompt and adjusting behavior for Opus 4.7. It includes a detailed rationale, measured prompt-size reductions, and install steps using tweakcc-fixed.

Eva is an open-source, AI-native coding agent from missingstudio that works as a terminal chat client and writes its conversation history to disk as it happens.

Graphite’s Five Percent research article analyzes writing-style differences in Claude Opus 5.5, compares it with prior Claude versions and GPT-6 Astra, and includes charts, tell counts, and downloadable data.

A repository of reusable AI agent skills centered on regression evals and benchmark history, with install instructions and local eval workflows.

A curated learning roadmap for building production AI agents, covering OpenAI, Anthropic, MCP, harnesses, evals, memory, and safety. The README organizes the material into phases with exercises and a prioritized reading table.

Anthropic research report on how people use Claude Code in practice, based on a large privacy-preserving analysis of session data. It covers task types, division of labor between user and model, and how domain expertise affects outcomes.

Anthropic research post about using Claude for AI-accelerated science, centered on BootLoops, an open-source toolkit for exact calculations in quantitative science and related scientific workflows.

A GitHub repository tracking vulnerabilities credited to the Anthropic research team, with a large visible list of CVEs, related findings, and submission guidance.

Anthropic’s Claude Platform docs page tracking system prompt updates for Claude web and mobile apps, with dated entries across model versions and a note that the updates do not apply to the API.

Anthropic help article explaining how Claude marks generated text with embedded watermarks and signed provenance metadata, plus what the markings cover, how to detect them, and key limitations.