A Digital Consulting Plus methodology
English
← Full mapCurated public summary
Index
Handbook
01DCP Flow: what it is and why it exists02How DCP Flow works03How to start a project04Kai and working with AI agents04AKai operational capabilities05Institutional Prompt Cookbook06Solution Discovery explained for humans07Architecture, stack, and technology ecosystem07ATool & MCP Registry07BHostinger MCP for DCP websites and hosting08Quality Gates, QA, security, and evidence08AQuality profiles and testing strategy09Git, branches, PRs, and multi-agent work10Deploy, release, support, and operational evolution10AContainer First and standardized host operations10BPractical Container First demo11How DCP Flow evolves12Real cases and lessons learned13How to teach DCP Flow14Practical DCP Flow use cases15End-to-end playbooks: Web and Laravel
Entry 07A43%
43%9 / 21
Entry 07APUBLIC HANDBOOK

Tool & MCP Registry

Operational tool/MCP registry with ADOPTED/EXPERIMENTAL/REJECTED/DEPRECATED/UNKNOWN states, guardrails, and relationships to Skills, Standards, and ADRs.

Source & authority

Public explanation

PUBLIC HANDBOOK
Based on
Human Handbook 07A — Tool & MCP Registry
Related normative sources
  • NORMATIVE SOURCEStandard · AI Project Context
Authority
Public explanation — NON-NORMATIVE
Summary

Operational tool/MCP registry with ADOPTED/EXPERIMENTAL/REJECTED/DEPRECATED/UNKNOWN states, guardrails, and relationships to Skills, Standards, and ADRs.

The Tool & MCP Registry prevents rediscovery and uncontrolled adoption. Every tool or MCP should have a purpose, lifecycle state, guardrails, and evidence appropriate to its maturity.

Registry states

ADOPTED marks regular approved use; EXPERIMENTAL requires a POC and limits; REJECTED records why a tool is unsuitable; DEPRECATED signals planned retirement; UNKNOWN means evidence is insufficient.

Key points
  • ADOPTED.
  • EXPERIMENTAL.
  • REJECTED.
  • DEPRECATED / UNKNOWN.

Relationship with DCP Flow

A tool never replaces a Standard, Blueprint, or Skill. The registry tracks what exists and how it is governed; reusable operating rules belong in the appropriate DCP Flow artifact when evidence supports promotion.

Key points
  • POC before promotion.
  • Guardrails and permissions.
  • Relationship to Skills and Standards.
  • ADR when the decision is material.

Knowing a tool does not mean adopting it

The Tool & MCP Registry preserves which tools are known, what they do, what risk they introduce, and what evidence exists. ADOPTED, EXPERIMENTAL, REJECTED, DEPRECATED, and UNKNOWN prevent a catalog mention from being interpreted as a universal recommendation. A tool adopted by DCP is also not automatically approved for every project or authorized to execute now.

Key points
  • KNOWN ≠ ADOPTED.
  • ADOPTED ≠ mandatory.
  • Project approval remains contextual.
  • Authorization remains a separate decision.

An MCP is a capability boundary

Before adopting an MCP, DCP Flow asks what it can read or mutate, how it authenticates, where it runs, what secrets it uses, what logs it produces, and which capability it duplicates. Registry v0.3 keeps GitHub MCP, n8n Instance MCP, Google Cloud MCP/gcloud-mcp, AWS MCP, and Hostinger API MCP as EXPERIMENTAL until an executable POC exists. Protocol convenience does not make the MCP a source of truth.

Key points
  • GitHub MCP — EXPERIMENTAL.
  • n8n MCP — EXPERIMENTAL.
  • GCP/AWS MCP — EXPERIMENTAL.
  • Hostinger MCP — EXPERIMENTAL.

Registry, Skill, Standard, and ADR have different responsibilities

The Registry records what tool is known and its status; a Skill explains how to execute a capability correctly; a Standard defines a cross-project rule; an ADR preserves a material contextual decision. Building a custom MCP only makes sense when the capability repeats, the contract is stable enough, guardrails are clear, and it improves meaningfully on existing APIs, CLIs, or MCPs.

Key points
  • Registry = inventory + status.
  • Skill = execution knowledge.
  • Standard = cross-project rule.
  • ADR = contextual material decision.
Contextual glossary＋
Tool Catalog

Governed registry of known tools and their status within DCP Flow.

MCP

Interface/protocol for exposing tools and resources to AI agents in a structured way.

ADOPTED

State of a tool validated and accepted for a defined use.

EXPERIMENTAL

State of a tool still under evaluation or POC.