Tag
This article proposes an opaque, interoperable passkey record format based on PHC strings and describes a Go API for handling them, aiming to simplify server-side passkey storage and library interoperability.
aiignore is an open specification and reference implementation for defining portable policies to control AI agents' access to files, environment variables, network, and output. The first public alpha release includes a TypeScript CLI and integrations for Codex and Gemini CLI.
Andrew Ng shares practical advice for building product specs when using AI coding agents for rapid prototyping, emphasizing that AI tokens are cheap while human tokens are valuable, and that iterating with agent-built prototypes is more efficient than upfront specification.
QUALITY.md is an open file format and CLI tool for defining and evaluating project quality, helping teams and agents align on quality standards.
This paper presents a normative specification for a data model called MMM, aimed at enabling knowledge interoperability in a decentralized knowledge commons.
The article discusses the incompatibility of AT URI syntax with standard URI/URL specifications due to placing DIDs in the authority field, and explores potential solutions and their trade-offs.
A technical specification for a robust job server tool hosted on Codeberg.
The x86 Ecosystem Advisory Group has published the AI Compute Extensions (ACE) specification, defining new x86 instructions and register state for accelerating matrix multiplication and reduced precision data formats in machine learning workloads.
The Agentic Resource Discovery Specification (ARD) defines a standard for AI clients to dynamically discover external capabilities such as tools, MCP servers, APIs, and other agents, enabling seamless integration beyond static knowledge.
Google has released the Open Knowledge Format, a specification based on Karpathy's llm-wiki, using markdown files for portable knowledge representation, decoupling human-written knowledge from agent consumption.
The article announces the launch of WASI 0.3, which integrates async primitives natively into WebAssembly components via the Component Model, simplifying APIs and enabling better component composition.
Google introduces DESIGN.md, a format specification that combines YAML front matter design tokens with Markdown prose to give coding agents a structured understanding of a design system.
A blog post detailing a development process that uses specialized AI subagents to systematically critique and improve specifications before implementation, aiming to rebuild trust in AI-assisted coding through automated doubt and multiple perspectives.
The Alliance for Open Media has released the AV2 v1.0.0 specification, a next-generation video coding standard offering superior compression efficiency for streaming, broadcasting, and real-time video conferencing.
Reed is a lightweight specification for implementing prefix authentication schemes (transparency logs) that produces shorter proofs than traditional certificate transparency logs, superseding the earlier Bamboo specification with greater efficiency and flexibility.
The MCP specification release candidate for 2026-07-28 introduces a stateless core, extensions like MCP Apps and Tasks, improved authorization, and a formal deprecation policy, enabling scalable HTTP infrastructure without sticky sessions.
The Procedural Fascicle for R7RS Large Scheme is a document that forms part of the R7RS Large standard for the Scheme programming language, intended to belong to the entire Scheme community and freely copied for use in manuals and documentation.
The MCP 2026-07-28 release candidate introduces a stateless core, extensions for server-rendered UIs and long-running tasks, improved auth, and a formal deprecation policy, simplifying deployment and scaling.
This article explores why adding assumptions to specifications or tests inherently weakens the resulting property, using logical implication and examples from formal methods and Rust. It also discusses practical reasons for using assumptions despite this weakening.
A Twitter user shares a prompt for AI code generation: when implementing a specification, keep a running implementation-notes file to document decisions, changes, and tradeoffs.