Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add sigistry/marketplace --skill mcp-2026-07-28git clone --depth 1 https://github.com/sigistry/marketplaceWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/sigistry/marketplace/mcp-2026-07-28)<a href="https://agentmods.dev/skills/sigistry/marketplace/mcp-2026-07-28"><img src="https://agentmods.dev/badge/skills/sigistry/marketplace/mcp-2026-07-28/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/sigistry/marketplace/mcp-2026-07-28"><img src="https://agentmods.dev/badge/skills/sigistry/marketplace/mcp-2026-07-28.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00128 | $0.01222 |
| Opus 5.5 | $0.00051 | $0.00489 |
| Sonnet 5.5 | $0.00026 | $0.00244 |
| Haiku 4.5 | $0.00013 | $0.00122 |
Grade A, and why
mcp-2026-07-28 scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 49 lines — stays where its author put it; the contents beside it link to each section on GitHub.
MCP 2026-07-28 Spec
Purpose
The MCP 2026-07-28 specification is the largest revision since the protocol launched. It moves MCP from a bidirectional, stateful protocol toward a stateless request/response core, so servers deploy on serverless and edge infrastructure and scale horizontally; it formalizes versioned extensions (MCP Apps for interactive UI, Tasks for long-running work) so capabilities can be added without touching the core; and it aligns authorization with production OAuth 2.1 / OIDC (covered in the mcp-authorization skill). This skill is the reference for the core, transport, and lifecycle so a server is shaped correctly, not merely functional.
What changed, at a glance
| Area | Before | 2026-07-28 |
|---|---|---|
| Transport | HTTP+SSE dual endpoint (GET /sse + POST /messages) |
Streamable HTTP, single endpoint; HTTP+SSE removed |
| Core model | Stateful, long-lived bidirectional session | Stateless request/response core; each request self-contained |
| Deployment | Sticky, single-instance | Serverless / edge / multi-replica behind a load balancer |
| Extensions | Ad-hoc | Versioned extensions framework (MCP Apps, Tasks) |
| Auth | Custom / bearer, loosely specified | OAuth 2.1 + OIDC aligned (see mcp-authorization) |
The stateless core
The single most important shape change: do not require per-connection server memory to serve a request. A request carries what it needs; any replica can answer it. Concretely:
- No in-process session map that a second instance would not share. If you must correlate requests, key off an external store (Redis, a DB) by a stable id, not a module-level
Map. - Long-running work is not a sticky session held open, it is the Tasks extension: start returns a task id, the client polls/subscribes, any replica can report status. See
references/versioned-extensions.md. - stdio servers are inherently single-process and per-user; the stateless rule is about remote (HTTP) servers that scale out.
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago First seen · 49 lines · 128 tokens per session scan A 5e26f88a456d
mcp-2026-07-28 is a skill published in the GitHub repository sigistry/marketplace (3 stars, last pushed 4d ago), licensed MIT. It adds 128 tokens to every session and 1,222 once invoked, about $0.0005 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.
Other skills, from other repositories
design-code-architecture
Guided journey from an app idea to a deliberate architecture: boundaries, domain model, data decisions, and resilience, making only the expensive-to-reverse decisions and deferring the rest. Orchestrates eight skills phase by phase - clean-architecture, domain-driven-design, system-design, ddia-systems…
system-design
Design scalable distributed systems using structured approaches for load balancing, caching, database scaling, and message queues. Use when the user mentions "system design", "scale this", "high availability", "rate limiter", "design a URL shortener", "design Twitter", "design Uber", "design a news feed", "system…
mcp-tool-dev
Create MCP tools — individual tool functions exposed via Model Context Protocol. Use this skill whenever users mention MCP tools, tool handlers, tool functions, tool definitions, or want to add capabilities to an MCP server. Also use when the conversation involves designing tool schemas, writing tool descriptions, or…
mcp-server-dev
Create MCP servers — multi-tool services exposed via Model Context Protocol. Use this skill whenever users mention MCP servers, building servers, server scaffolding, tool registration, or want to package tools for distribution via uvx or npx. Also use when the conversation involves stdio transport, server lifecycle…
architect
This skill should be used when the user asks to "design system architecture", "evaluate microservices vs monolith", "create architecture diagrams", "analyze dependencies", "choose a database", "plan for scalability", "make technical decisions", or "review system design".
backend
This skill should be used when the user asks to "design REST APIs", "optimize database queries", "implement authentication", "build microservices", "review backend code", "set up GraphQL", "handle database migrations", or "load test APIs".