Iron Cannon
An MCP-powered compliance copilot for SaaS stacks, enabling structured audit workflows including stack detection, module wiremapping, implementation directives, code verification, and security/infrastructure/legal readiness gates.
README
Iron Cannon
An MCP-powered compliance copilot for SaaS stacks — especially Next.js / Cloudflare Workers + D1 + Stripe + Resend.
Iron Cannon gives AI agents (and humans) a structured audit workflow: detect the stack, generate a module wiremap, ship implementation directives, verify code patterns, and gate production / infrastructure / legal readiness. It runs locally as a Cloudflare Worker dev server and exposes tools over the Model Context Protocol (HTTP JSON-RPC).
Portfolio note: This project was dogfooded on real Cloudflare + Stripe apps but is paused before production SaaS launch. It is a strong demonstration of MCP tool design, rules engines, and agent guardrails — not a hosted compliance certification product.
What it does
| Tier | Tools | Purpose |
|---|---|---|
| Pro | T01–T08 | Stack discovery, wiremap, module directives, snippet verification, drift & recovery |
| Armor | T09–T11 | Security surfaces, infra checklists (cache, rate limits, stress testing) |
| IronClad | T12–T14 | Legal obligation map + spot-check verification |
Design philosophy: Iron Cannon is two modes in one product:
- Copilot — rich checklists and sequencing (high trust)
- Judge — pass/fail gates only after deliberate code snippets and verified modules (medium trust)
Architecture
┌─────────────────┐ HTTP /mcp ┌──────────────────────┐
│ Cursor / CLI │ ◄──────────────► │ apps/mcp-worker │
│ (MCP client) │ JSON-RPC 2.0 │ (Cloudflare Worker) │
└─────────────────┘ └──────────┬───────────┘
│
┌──────────▼───────────┐
│ @ironcannon/mcp-core │
│ T01–T14 tool runtime │
└──────────┬───────────┘
│
┌─────────────────────────────────┼─────────────────────────┐
│ │ │
@ironcannon/compose @ironcannon/compare @ironcannon/verify
(tier redaction) (pattern matching) (stack completeness)
Rules engine data lives under docs/engine/ as JSON specimens (fixtures, obligations, catalogs). npm run build:worker-bundle packs them into packages/mcp-core/src/generated/ for the Worker.
Quick start (local MCP)
Prerequisites
- Node.js 20+
- npm
1. Install & build
git clone https://github.com/DillonRich/Iron_Cannon.git
cd Iron_Cannon
npm install
npm run build:worker-bundle
2. Verify the engine
npm run g2:audit # package + smoke tests (HTTP optional)
npm run portfolio:preflight # scan for accidental live secrets
3. Start the MCP server
npm run ironcannon:serve
Server listens at http://127.0.0.1:8787 (/health, /mcp, /tools).
4. Connect Cursor (or any MCP client)
Copy mcp-config.example.json into your Cursor MCP settings and adjust paths per Cursor docs. Use one server URL with different x-ironcannon-tier headers (pro, armor, ironclad).
Toggle the MCP server off/on after starting ironcannon:serve.
CLI workflows
# Audit an external project (T01–T03 local, prints wiremapAttestation)
npm run ironcannon -- audit "C:/path/to/your-app"
# Golden path smoke (reference app)
npm run ironcannon -- ref-stack
npm run ironcannon -- ref-golden
# Full tiered harness (Pro → Armor → IronClad)
npm run ironcannon -- full
# Dogfood bundle: audit + fixture regression + MCP envelope test
npm run ironcannon:dogfood -- "C:/path/to/your-app"
# Single module verify
npm run ironcannon -- verify M12-stripe-webhook ./path/to/snippet.ts
Set tier: IRON_CANNON_TIER=armor npm run ironcannon -- ...
Agent workflow (summary)
- T01
analyze_project_stack— detect frontend, compute, DB, Stripe/Resend - T02
validate_stack_completeness— missing secrets / config hints - T03
generate_system_wiremap— module chain +wiremapAttestationtoken - T04
get_module_directives— per-module implementation guidance (+snippetHintfor verify-heavy modules) - T05
verify_module_compliance— must pass real code snippets (notAuditedif empty) - Armor: T09–T11 security + infrastructure gates
- IronClad: T12–T14 legal obligation map (T13 optional snippet verify)
Pass wiremapAttestation from T03 on all T04/T05/T11 calls.
Project layout
| Path | Description |
|---|---|
packages/mcp-core/ |
Core MCP tool runtime (T01–T14) |
packages/compose/ |
Tier-based response redaction |
packages/compare/ |
Pattern / obligation compare engine |
packages/verify/ |
Stack completeness checks |
apps/mcp-worker/ |
Cloudflare Worker HTTP surface |
scripts/ |
CLI (ironcannon.mjs), build, test harnesses |
docs/engine/ |
Rules JSON, fixtures, obligation index (not markdown) |
examples/golden-reference-app/ |
Minimal reference stack for T01–T05 |
Testing
npm run g2:audit # recommended smoke gate
npm run planning:guardian-retest # pattern equivalence fixtures
npm run g2:golden-full # full Pro + Armor + IronClad path
npm run planning:user-journey # 121 behavioral scenarios (slower)
Known limitations (honest)
- Stack coverage is strongest for Cloudflare Workers/Pages + Stripe + Resend — not universal.
- T05 / T11 judge mode requires deliberate snippets and
verifiedModules— not unattended certification. - Hosted deploy (Cloudflare production, D1 platform DB, Vectorize) is scaffolded but not required for local MCP.
- Legal (IronClad) outputs checklists and heuristics — not legal advice.
Security
See SECURITY.md. Dev-only API keys in docs/engine/phase1/fixtures/api-keys.dev.json are fake local tokens (ic_dev_*). Do not commit real .env files.
License
GNU AGPL v3 — see LICENSE. You may run this locally for audits and contribute improvements; if you offer this software as a network service, AGPL copyleft applies.
Author
Built as a portfolio project demonstrating MCP server design, structured agent guardrails, and compliance-oriented tooling for modern SaaS infrastructure.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.