Security reviewer from a repository skill

Why a skill

Your team may already keep its security rules as an agent skill, for coding assistants. A custom agent can follow the same skill in code review: it opens the skill when the change needs it, reads its reference files, and quotes the rule each finding applies. When the skill changes on the main branch, the agent follows the new version with no settings change. The example is a multi-tenant scheduling API where every query must be scoped to the caller’s workspace.

Step 1 — The skill in the repository

.claude/skills/tenant-isolation/SKILL.md

---
name: tenant-isolation
description: Rules that keep one workspace's data out of another's. Use when changing endpoints, queries or background jobs.
---

# Tenant isolation

- Every query on a tenant table filters by `workspaceId` taken from the authenticated session, never from the request body or query string.
- Every new endpoint declares `@RequireWorkspaceRole(...)`; an endpoint without it is public.
- Background jobs receive `workspaceId` in their payload and pass it to every repository call.
- Logs never include tokens, passwords or full email addresses.

For the list of tenant tables, see references/tenant-tables.md.

.claude/skills/tenant-isolation/references/tenant-tables.md

# Tenant tables

bookings, calendars, invoices, members, resources

The folder name and name must match, and description is required.

Step 2 — Import it as an agent

Open the repository's settings

Pick the repository in the scope picker, then What to review → Custom agents → New agent.

Import from repository

The tab lists the skills found on the default branch. Click tenant-isolation.

Check the result

The agent is created off, named Tenant isolation reviewer, with the skill linked and these instructions:

Review the change against the team's "tenant-isolation" skill: load it, follow its rules and references, and report where the change departs from them. Cite the rule you apply.

Step 3 — Sharpen the instructions

Keep the skill reference and add what proof a finding needs:

Instructions

Review the change against the team's "tenant-isolation" skill: load it, follow its rules and references, and report where the change departs from them. Cite the rule you apply.

Trace where workspaceId comes from in every query or repository call the change adds or edits. A finding must show the line where the value enters (session, body, query string, job payload) and the line where it is used.

Ignore tests and seed scripts. Do not report general security issues the skill does not cover; Kody's own review handles those.

Field	Value
Files	`src/api/`, `src/jobs/`, `src/repositories/**`
Except	`*/.test.ts`, `src/db/seeds/**`

Step 4 — Optional: a plugin

If your issue tracker is installed as a plugin, turn it on under Tools → Plugins. Only read tools are offered. When a pull request description links a ticket, the agent can read it and check the change against what it asks.

Step 5 — Comments

Field	Value	Why
Format	Inline comments
Minimum severity	High	Only tenant leaks and missing auth, not hardening ideas.
Suggested fix	When confident

Click Save.

Step 6 — Try it

Run Try it on a pull request that adds an endpoint. Check:

Skills opened lists .claude/skills/tenant-isolation, and references/tenant-tables.md when the agent needed it.
Kept findings quote a rule from the skill:

**Rule** (`.claude/skills/tenant-isolation/SKILL.md`):
> Every new endpoint declares `@RequireWorkspaceRole(...)`; an endpoint without it is public.

Dropped findings show their reason. A general security remark the skill doesn’t cover is dropped as outside the agent’s focus, which is what you want here.

Turn the agent on. To get a second opinion on an open pull request later, comment @kody -v tenant-isolation-reviewer.

Checklist

SKILL.md has name matching its folder and a description
Reference files sit one folder below the skill at most
Instructions say what proof a finding needs
Minimum severity matches how loud the agent should be
Tried on a pull request with a known gap

See Skills & docs for detection and budgets.

Custom Plugins

Custom Context Playbooks

Recipes

Custom Agents

LLM Provider Setup

Security reviewer from a repository skill

Why a skill

Step 1 — The skill in the repository

Step 2 — Import it as an agent

Step 3 — Sharpen the instructions

Step 4 — Optional: a plugin

Step 5 — Comments

Step 6 — Try it

Checklist

​Why a skill

​Step 1 — The skill in the repository

​Step 2 — Import it as an agent

​Step 3 — Sharpen the instructions

​Step 4 — Optional: a plugin

​Step 5 — Comments

​Step 6 — Try it

​Checklist

Why a skill

Step 1 — The skill in the repository

Step 2 — Import it as an agent

Step 3 — Sharpen the instructions

Step 4 — Optional: a plugin

Step 5 — Comments

Step 6 — Try it

Checklist