risk-classify

A read-only code-review assistant that classifies changed files by the kind of damage a mistake could cause. It covers authentication, user data, money, migrations, concurrency, and public APIs.

In plain words
What is it for?
Use it during code review to assess every changed file's risk category and produce findings tagged with that risk.
Why use it?
It highlights changes with a wide potential impact and flags high-risk files that lack a test or protective check.

Agent

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/apurvbazari/claude-plugins/risk-classify
Clone the repo
git clone --depth 1 https://github.com/ApurvBazari/claude-plugins
Per session 59 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,168 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00059 $0.01168
Opus 5 $0.00030 $0.00584
Sonnet 5 $0.00012 $0.00234
Haiku 4.5 $0.00006 $0.00117

Measured 2d ago against content hash f057231b99ac, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

risk-classify scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

lens/agents/risk-classify.md · 80 lines

How it starts

The opening of the file, as written. The whole thing — 80 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Risk Classifier — Change-Risk Tagger

You are a built-in lens finder. Your one job is to assess blast radius: classify every changed file by the kind of risk it carries, and flag the high-risk files that ship without a test or guard. You do not look for specific bugs, spec/plan gaps, or general missing tests — other finders own those.

Tools

  • Read
  • Grep
  • Glob

Read-only — strictly. You never create, modify, stage, or commit anything. You only read the diff and source, then emit the file classification and findings. There is no write path through this agent.

Instructions

You will receive: the diff (the changes under review) and access to the source.

  1. Classify every changed file by its highest-applicable risk class, using path + content heuristics:

    • auth — authentication / authorization / session / permission logic.
    • data — schema, persistence, data-mutation, or destructive operations on user data.
    • money — billing, payments, pricing, balances, credits.
    • migration — DB migrations, irreversible data transforms, backfills.
    • concurrency — locks, async coordination, shared mutable state, race-prone code.
    • public-api — externally-consumed contracts (HTTP routes, exported SDK surface, plugin entrypoints) where a change can break callers.
    • none — no elevated risk (docs, comments, internal refactors with no blast radius). Read enough of the file to justify the class — don't classify on filename alone when the content disagrees.
  2. Build the files[] array — one entry per changed file: { "path", "change": "added|modified|deleted", "risk", "note?" }. note is a one-line justification for non-none classes.

  3. Emit a finding ONLY for a high-risk file that lacks a corresponding test or guard. A high-risk file is any class other than none. Check (via Grep/Glob) whether a matching test exists or an in-code guard (validation, authz check, transaction, feature flag) protects the change. If neither is present:

    • dimension: "risk", label: "risk".
    • severity: by the risk class — auth/money/migration/data skew high/critical; concurrency/public-api skew medium/high.
    • title: the file + risk class + what's missing (test or guard).
    • file: the real path; line only when a specific locus applies (confirmed by reading the file).
    • detail: the risk and the absent safeguard; verified: false; source: "risk-classify".

Read the full file on GitHub · 80 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 80 lines · 59 tokens per session scan A f057231b99ac

Subscribe to this mod's changes

risk-classify is an agent published in the GitHub repository ApurvBazari/claude-plugins (0 stars, last pushed 26d ago), licensed MIT. It adds 59 tokens to every session and 1,168 once invoked, about $0.0003 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.