Skip to content
View snapsynapse's full-sized avatar

Highlights

  • Pro

Block or report snapsynapse

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
snapsynapse/README.md

Sam Rogers

Building open protocols and measurement infrastructure for the agentic web and for calibrated AI Collaboration.

Founder, PAICE.work · Snap Synapse

Systems don't generally fail because the tech is bad. They fail at the handoffs between roles, between agents, between decisions that no one designed to connect. I build the standards, diagnostics, and tooling that make human-AI collaboration structurally sound, not just technically possible.

Interoperability over integration · Measurement over intuition · Structure over speed


Work

Open standards and tooling released independently because interoperability is the point, not lock-in.

Protocols & Standards

  • graceful-boundaries — How services communicate operational limits and error context to humans and autonomous agents
  • turnfile — File-based coordination for LLM agents collaborating on shared codebases without real-time communication
  • hardguard25 — Human-friendly 25-character alphabet for unambiguous IDs

Agent Tooling

  • skill-provenance — Version tracking for agent skill bundles across sessions, surfaces, and platforms
  • Siteline — Grades whether a website presents a usable path for AI agents acting on behalf of humans
  • skill-a11y-audit — Drop-in WCAG 2.1 AA accessibility audit for AI coding agents

Knowledge Infrastructure

  • Knowledge-as-Code — Pattern for human & machine-readable structured knowledge bases via database-less static sites, JSON APIs, and multi-output from markdown
  • EveryAILaw.com — Which AI regulations you care about, and why. Obligation-first reference for compliance teams and AI agents
  • AITool.watch — What AI tools can actually do: capabilities, pricing, and platform support across major providers

Companies

  • PAICE.work PBC measures how effectively people and AI systems collaborate. Not whether teams use AI, but whether the collaboration is actually working. Behavioral diagnostics for People + AI Collaboration Effectiveness.
  • Snap Synapse LLC helps teams and orgs apply AI responsibly, without the usual drama or damage. Structured implementations that move from AI promise to AI practice.

Collaborating

Currently looking for people working on agent communication protocols, human capability measures, AI regulations & governance structures, and aggregate intelligence infrastructure. If that's you, please reach out: hello @ sam-rogers .com


Elsewhere

Pinned Loading

  1. graceful-boundaries graceful-boundaries Public

    A specification for how services communicate operational limits and error context to humans and autonomous agents.

    4

  2. skill-provenance skill-provenance Public

    Finally a source of truth for agent skill bundles! Use this metaskill for version tracking across sessions, surfaces, and platforms. Tracks identity, staleness, and intent inside the bundle.

    Shell 3 1

  3. ai-tool-watch ai-tool-watch Public

    Evidence-backed reference for AI capabilities, pricing tiers, availability, and platform support across ChatGPT, Claude, Gemini, Copilot, and more.

    JavaScript 7

  4. turnfile turnfile Public

    file-based protocol for making LLM agents collaborate on shared codebases without real-time communication

    JavaScript 5

  5. skill-a11y-audit skill-a11y-audit Public

    Drop-in accessibility audit for AI coding agents. WCAG 2.1 AA scanning with template-aware sampling, remediation hints, and progress tracking.

    JavaScript 1

  6. hardguard25 hardguard25 Public

    A 25-character alphabet for human-friendly unique IDs. Like Crockford Base32, but removes 11 confusable characters instead of 4.

    JavaScript