Support the Open Infrastructure

These open-source projects are free and always will be. Sponsorship keeps specs evolving, tests passing, and tooling free for everyone.

All projects share the same tiers:

Sponsorship does not buy: SLAs, roadmap control, private builds, or consulting (available separately via Snap Synapse).

Graceful Boundaries

The spec that teaches services to say "not right now" instead of crashing.

Defines how services communicate operational limits to humans and agents across four conformance levels. Reference implementation powers Siteline. 131+ tests. No existing standard covers this.

Sponsors: API platform teams, agent framework builders, agentic-web infrastructure.

Skill Provenance

Know where your agent's skills came from, and whether they've changed.

Version identity and manifest tracking that travels with skill bundles across Claude, Codex, and other platforms. Supply-chain security for the agent ecosystem. 22 core + 4 distribution evals.

Sponsors: AI safety teams, agent platform builders, enterprises deploying custom skills.

Turnfile

The protocol for AI agents that disagree productively.

Structured turn-taking for multi-agent systems that need to negotiate and converge without a boss agent or voting. Tested across 11+ real Claude/Codex sessions.

Sponsors: Multi-agent system builders, AI governance researchers, decentralized AI workflow teams.

Knowledge-as-Code Template

The build system behind three live knowledge references.

Reusable architecture for structured, version-controlled knowledge bases. Ontology-first, zero dependencies, Git-native. Define your domain as YAML, get HTML sites, JSON APIs, MCP servers, and discovery files from one source. Powers AI Tool Watch, Every AI Law, and Virtual Classroom Watch.

Sponsors: Documentation teams, developer tool builders, anyone outgrowing a wiki.

AI Tool Watch

Plain-English AI capability tracking, verified by four models twice a week.

Tracks capabilities, plans, and constraints across 12 major AI products. Every claim verified through a four-model consensus cascade (Gemini, Perplexity, Grok, Claude). Live at aitool.watch with 87 implementations and 125 bridge pages. WCAG 2.1 AA compliant.

Sponsors: AI educators, product managers, enterprises making platform decisions, training facilitators.

skill-a11y-audit

Drop-in accessibility audits for AI coding agents.

Portable skill giving any agent (Claude, Codex, etc.) the ability to run WCAG 2.1 AA audits, combining axe-core and Lighthouse with compliance mapping and structured reporting. Most agents can write code; few can audit it.

Sponsors: Web agencies, design systems teams, accessibility consultancies, compliance-driven enterprises.