Skip to content

CodeGate Guardian

CodeGate Guardian is an AI Security Posture Management platform for the age of agentic development.

Code Video Cerebral Valley

Core trick

A local CLI inventories AI tool configs and pushes reports to a server that receives and evaluates them but never sends commands back.

Tags

securitydev tools

Description

CodeGate Guardian is an AI Security Posture Management platform for the age of agentic development. It scans GitHub repositories and AI-tool artefacts (skills, MCP configs, hooks, rules, and agent instructions), detects high-risk behaviours, and transforms raw findings into clear governance reporting. In one workflow, teams can see risk posture, policy compliance, asset coverage, and ownership accountability across projects, helping security and engineering leaders identify issues early and drive remediation with confidence.
jonathansantilli/codegate-guardian★ 10 forksTypeScriptMITCreated Mar 21, 2026 · last push Oct 19, 2026

README excerpt

# CodeGate Guardian

Guardian is a self-hosted console for seeing what AI tooling your developers
actually have installed, and which of it is dangerous.

The `codegate` CLI runs on each developer machine. It inventories every AI
tool, skill, MCP server and rules file it finds, scans them, and reports what
it saw to this server. Guardian aggregates those reports and shows them.

**This server never sends anything to a machine.** It receives, evaluates and
displays. Remediation happens on the machine, by the person who owns it, and
the next report is the evidence it happened. Revoking an enrolment stops this
server accepting a machine's reports; it does not reach the machine.

## What it shows

- **Machines** — every machine reporting, who is accountable for it, what it
  carries, and when it last checked in.
- **Inventory** — artifacts keyed by **content hash, not by name**. Two files
  sharing a name but differing by one byte are two artifacts, so a malicious
  skill cannot hide behind a familiar filename.
- **Findings** — with a lifecycle nobody has to maintain: a finding is open
  while a machine still reports it, and resolves when a later report from that
  machine no longer contains it. Status is derived, never stored.
- **Policies** — rules evaluated here against what each machine reported.
  Guardian flags a violation; it cannot block anything on a laptop.
- **Activity** — who or what did something, and the API call behind it.
- **API & access** — the console is a clien

Team