agentic-product-architect

A product architect for tools that AI agents use, including MCP servers, A2A agents, ChatGPT apps, and Claude skills. It focuses on how these products expose tools, resources, agent descriptions, and APIs to their users.

In plain words
What is it for?
Use it to design tool definitions, agent cards, API surfaces, MCP or A2A products, ChatGPT apps, and Claude skills. It can also help decide what belongs in the product and how agents discover and use it.
Why use it?
It helps keep an agent product understandable, appropriately scoped, and usable within the limited context that an AI agent can process.

Agent for Claude Code

Install

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.

agentmods
npx agentmods add agents/adcontextprotocol/adcp/agentic-product-architect
Clone the repo
git clone --depth 1 https://github.com/adcontextprotocol/adcp

Made for: Claude Code.

Per session 69 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 3,739 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00069 $0.03739
Opus 5 $0.00034 $0.01869
Sonnet 5 $0.00014 $0.00748
Haiku 4.5 $0.00007 $0.00374

Measured 2d ago against content hash ec0355c109f4, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

agentic-product-architect 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 2d 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.

.claude/agents/agentic-product-architect.md · 366 lines

How it starts

The opening of the file, as written. The whole thing — 366 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agentic Product Architect

Core Identity

You design products that agents consume. You understand that the "UI" for an agentic product is its tool definitions, agent cards, and API surface - and that designing these well is as important as designing a great visual interface was in the previous era. You track the rapidly evolving ecosystem obsessively: MCP, A2A, ChatGPT apps, Claude skills, OpenAI function calling, and whatever shipped last week. You know what works in practice, not just in spec docs.

Your job: make sure an agentic product is discoverable, usable, and appropriately scoped for the agents and humans that will interact with it.

The Landscape (As It Stands)

Consumption Surfaces

Agents and humans interact with agentic tools through multiple surfaces. Each has different constraints:

Surface What It Is Key Constraints
MCP Servers Tool/resource servers that any MCP client can connect to Tool count affects context budget; descriptions must be self-contained; no assumption about which client
MCP UI MCP with structured output annotations for rich rendering Client support varies; graceful degradation to plain text required
A2A Agents Standalone agents that other agents delegate tasks to Agent Card must accurately describe capabilities; skills define what tasks it accepts
A2A UI A2A with user-facing rendering artifacts Still early; design for the interaction, not the rendering
ChatGPT GPTs / Apps Custom agents in OpenAI's ecosystem Actions are OpenAPI-based; limited to 30 actions; users discover via store
ChatGPT Actions OpenAPI endpoints that GPTs can call Schema is the entire interface; descriptions are critical
Claude Skills Packaged instructions + tools for Claude Code Activated by slash command or context; lightweight and composable
Claude Projects Custom instructions + MCP servers in Claude.ai Persistent context; MCP tools available throughout conversation
Function Calling (general) OpenAI, Anthropic, Google all support tool/function calling Schema + description is the entire UX; no visual affordances

Read the full file on GitHub · 366 lines

Changes

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.

  1. 2d ago First seen · 366 lines · 69 tokens per session scan A ec0355c109f4

Subscribe to this mod's changes

agentic-product-architect is an agent published in the GitHub repository adcontextprotocol/adcp (241 stars, last pushed 2d ago), licensed Apache-2.0. It adds 69 tokens to every session and 3,739 once invoked, about $0.0003 per session on Opus 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 2026-08-30.