MCP App Store
Developer Tools
ArmorCodex icon

ArmorCodex

by ArmorIQ AI

Open in ChatGPT

Overview

ArmorCodex lets you read, update, and review the policies that govern how AI coding agents interact with tools, files, and infrastructure. Set rules in natural language ("deny webfetch", "require approval for production database changes"), check current policy state, and register intent plans before tool calls run. Built by ArmorIQ for teams that need governance and audit on AI agent activity.

Tools

policy_read

ChatGPT
Returns the current ArmorCodex policy rule set held in this server's in-memory store. With no arguments, returns the full policy object including all rules and the current version. With an id argument, returns the single rule matching that id. Annotations: readOnlyHint=true because the tool only reads server state without modifying anything; destructiveHint=false because no state changes; idempotentHint=true because repeated calls with the same input return the same result without side effects; openWorldHint=false because all reads come from this MCP server's memory and no external system is contacted.

policy_read

ChatGPT
Returns the current ArmorCodex policy rule set held in this server's in-memory store. With no arguments, returns the full policy object including all rules and the current version. With an id argument, returns the single rule matching that id. Annotations: readOnlyHint=true because the tool only reads server state without modifying anything; destructiveHint=false because no state changes; idempotentHint=true because repeated calls with the same input return the same result without side effects; openWorldHint=false because all reads come from this MCP server's memory and no external system is contacted.

policy_update

ChatGPT
Updates the ArmorCodex policy rule set held in this server's in-memory store. Supports two modes: merge (default) upserts the submitted rules into the existing set, and replace overwrites the entire rule set with the submitted payload. The previous rules are not snapshotted, so replace cannot be undone from this tool. Annotations: readOnlyHint=false because the tool mutates server state; destructiveHint=true because replace mode irreversibly drops the prior rule set; openWorldHint=false because all state lives in this MCP server's memory and no external system is contacted.

policy_update

ChatGPT
Updates the ArmorCodex policy rule set held in this server's in-memory store. Supports two modes: merge (default) upserts the submitted rules into the existing set, and replace overwrites the entire rule set with the submitted payload. The previous rules are not snapshotted, so replace cannot be undone from this tool. Annotations: readOnlyHint=false because the tool mutates server state; destructiveHint=true because replace mode irreversibly drops the prior rule set; openWorldHint=false because all state lives in this MCP server's memory and no external system is contacted.

register_intent_plan

ChatGPT
Records a structured intent plan (goal plus an ordered list of steps describing the tools the agent intends to call) into this server's in-memory plan list and returns the generated plan id. This hosted server stores the plan only; it does not enforce or block subsequent tool calls. The companion local Codex plugin uses these plans to gate tool execution via PreToolUse hooks on the user's machine, but that enforcement is out of scope for this hosted MCP. Annotations: readOnlyHint=false because the tool appends a new record to the in-memory plan list; destructiveHint=false because each call appends a new plan and never modifies or deletes prior plans; idempotentHint=false because calling the tool twice with identical inputs produces two distinct plan ids and two records; openWorldHint=false because all state lives in this MCP server's memory and no external system is contacted.

register_intent_plan

ChatGPT
Records a structured intent plan (goal plus an ordered list of steps describing the tools the agent intends to call) into this server's in-memory plan list and returns the generated plan id. This hosted server stores the plan only; it does not enforce or block subsequent tool calls. The companion local Codex plugin uses these plans to gate tool execution via PreToolUse hooks on the user's machine, but that enforcement is out of scope for this hosted MCP. Annotations: readOnlyHint=false because the tool appends a new record to the in-memory plan list; destructiveHint=false because each call appends a new plan and never modifies or deletes prior plans; idempotentHint=false because calling the tool twice with identical inputs produces two distinct plan ids and two records; openWorldHint=false because all state lives in this MCP server's memory and no external system is contacted.

Capabilities

Writes

App Stats

6

Tools

ChatGPT

Platforms

Category

Technical Knowledge Search

Works with

ChatGPT

Similar Apps

Desktop Commander icon

Desktop Commander

Data Dolphin icon

Data Dolphin

KARP Inspector Lite icon

KARP Inspector Lite

IcePanel icon

IcePanel

Git Diff Patcher Bridge icon

Git Diff Patcher Bridge

Visily icon

Visily

Data refreshed daily