nodefony-rfc

Cites and applies IETF RFCs, W3C specs, MCP, and AGENTS.md conventions from raw sources.

Updated Dec 19, 2023
One-click install
npx skills add https://github.com/nodefony/nodefony-core --skill nodefony-rfc-nodefony
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: nodefony-rfc
Source: https://github.com/nodefony/nodefony-core/tree/main/.claude/skills/nodefony-rfc
Command: npx skills add https://github.com/nodefony/nodefony-core --skill nodefony-rfc-nodefony

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve? When implementing HTTP, WebSocket, cookies, CORS, OAuth, or MCP behavior in the Nodefony framework, developers risk inventing plausible-but-wrong behavior or citing outdated living specifications. This Skill provides the canonical normative sources — raw RFC texts, a frozen offline copy of MCP revision 2026-07-28, and the AGENTS.md convention — plus a drift-detection script that reports when a frozen spec has diverged from upstream. ## Core Features & Use Cases - Canonical source routing: Maps each functional area to its governing document — RFC 9110 (HTTP semantics), RFC 9113 (HTTP/2), RFC 6455 (WebSocket), RFC 6265 (cookies/SameSite), WHATWG Fetch (CORS), and OAuth RFCs 9728/6750/8707/8414. - Offline MCP 2026-07-28 mirror: The full specification tree, TypeScript schema, JSON schema, and per-message examples are frozen under references/mcp-2026-07-28/, readable with zero network requests. - AGENTS.md convention facts: Frozen upstream files and verified quotes covering precedence rules, editor size limits (Codex 32 KiB truncation), and the distinction between .agents/ proposals and the de facto .agents/skills/ root. - Drift detection: scripts/check-amont.mjs compares frozen copies against upstream GitHub SHAs and names exactly which tracked files changed (exit 0 = current, 3 = drift, 78 = unknown). - Use Case: While implementing server/discover for a Nodefony MCP server, consult the frozen spec and schema to confirm the required resultType field and error codes -32020/-32022 instead of guessing from memory. ## Quick Start Ask the agent which RFC or spec section governs the HTTP status code, WebSocket frame, cookie attribute, or MCP message you are implementing, and have it extract only the relevant section before writing code.

Frequently Asked Questions about nodefony-rfc

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I check which RFC governs an HTTP status code or header?

HTTP semantics including status codes, headers, and methods are defined by RFC 9110, which replaces RFC 7231. Fetch the plain-text version from ietf.org and grep for the exact keyword such as "401 Unauthorized" rather than reading the full document.

How do I implement server/discover for MCP revision 2026-07-28?

Servers MUST implement server/discover to advertise supported protocol versions, capabilities, and identity. The frozen offline copy under references/mcp-2026-07-28 contains the spec page, the TypeScript schema, and canonical JSON examples to validate your output against.

Does the skill include the full text of OAuth RFCs like 6750 and 8707?

No. Full RFC texts live in a single shared corpus under the nodefony-framework-dev skill to avoid duplicate unsynchronized copies. This skill explains what each RFC decides and points to that corpus for grep-based reading.

How do I know if a frozen spec copy has drifted from upstream?

Run node scripts/check-amont.mjs. It compares each references folder's AMONT.json SHA against the upstream GitHub repository, exiting 0 when current, 3 when tracked files changed, and 78 when the network check is inconclusive.

Is there a size limit for AGENTS.md files?

The AGENTS.md convention itself defines no size limit, but editors impose caps. OpenAI Codex concatenates AGENTS.md files from root to the current directory and silently truncates at 32 KiB, while Cursor recommends under 500 lines and Claude Code under 200.

Why should I avoid loading entire RFC documents into context?

RFCs are hundreds of pages long and loading them wastes tokens on irrelevant content. The recommended pattern is curl piped to grep with a targeted keyword and context flag, extracting only the section needed, then adapting the syntax rather than copying verbatim.