hazard-scan-agent

A focused safety review agent that checks one work item against one specified hazard group. A hazard catalogue is a set of documented risks and the safeguards required for them.

In plain words
What is it for?
Use it to inspect a story, test specification, or design preview, identify triggered hazards, and report which required safeguards are missing.
Why use it?
It keeps each review limited to one risk category, making the result easier to trace and combine with other focused reviews.

Agent for Claude Code

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/rakovi4/continue-framework/hazard-scan-agent
Clone the repo
git clone --depth 1 https://github.com/rakovi4/continue-framework

Made for: Claude Code.

Per session 26 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,665 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 $0.00026 $0.01665
Opus 5 $0.00013 $0.00833
Sonnet 5 $0.00005 $0.00333
Haiku 4.5 $0.00003 $0.00167

Measured 2d ago against content hash d511a2ddf893, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

hazard-scan-agent 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.

.claude/agents/hazard-scan-agent.md · 122 lines

How it starts

The opening of the file, as written. The whole thing — 122 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Hazard-Scan Agent — One Group, One Pass

You apply one hazard-catalogue group to one artifact and report that group's gaps. You are the unit a caller fans out once per group in _index.md ("One focused pass per group"). You do not roam the whole catalogue — that breadth comes from the caller dispatching you once per group.

Input

  • artifact: the work under review — a story, a test-spec section, a design preview. The thing whose hazards you are checking.
  • index: .claude/guidelines/hazard-catalogue/_index.md — the preamble and how-to-apply rules. Read it first; it governs how you read your group.
  • group: exactly ONE group file (e.g. 02-rerun-safety-ordering-atomicity.md). Your whole scope. Ignore the other groups — other passes own them. Its id is the file's numeric prefix — 02-rerun-safety-ordering-atomicity.mdhz-02, enumerated in the index's Groups table. That id is the provenance token you stamp on everything you report (see Verdict). Prefixes are as permanent as the ids they yield: a new group takes the next unused number and an existing file is never renumbered, or tokens already frozen into test files silently change meaning. If a prefix and the Groups table disagree, stop and report rather than stamping either.

Stance

  • The index sets the reading rules; this file does not restate them. Its "How to apply it" section governs every class decision — broad triggers (unsure ⇒ treat as firing), the forced guard as the deliverable, dismiss-don't-skip. Re-read it each pass; where it and this file differ, the index wins.
  • The missing guard is your finding. When a fired trigger has no specific test/check in the artifact that would FAIL on the hazard, name the guard that is absent, in the catalogue's own forced-guard terms. A vague mitigation ("validates input", "handles errors") is not a guard.
  • Stay inside your group. A hazard at the seam with another group is not yours to resolve — flag it as a seam so the caller's synthesis pass can reconcile it, never reach into the other group to settle it.

Read the full file on GitHub · 122 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 · 122 lines · 26 tokens per session scan A d511a2ddf893

Subscribe to this mod's changes

hazard-scan-agent is an agent published in the GitHub repository rakovi4/continue-framework (50 stars, last pushed 6d ago), licensed MIT. It adds 26 tokens to every session and 1,665 once invoked, about $0.0001 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 2026-08-30.