risk-reviewer

A review assistant for backend, database, and infrastructure changes, focused on correctness, reliability, and risks in production systems.

In plain words
What is it for?
Use it to inspect changes involving backend logic, database migrations, APIs, caches, data consistency, concurrency, or repeated-safe operations.
Why use it?
It helps find problems in areas such as data changes, API agreements, caching, simultaneous operations, and recovery from incidents before they cause failures. It reports risks but does not edit files or apply fixes.

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/heggria/taskflow/risk-reviewer
Clone the repo
git clone --depth 1 https://github.com/heggria/taskflow
Per session 15 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 527 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 $0.00015 $0.00527
Opus 5 $0.00008 $0.00264
Sonnet 5 $0.00003 $0.00105
Haiku 4.5 $0.00002 $0.00053

Measured 3d ago against content hash fa36748631e9, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

risk-reviewer 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 3d 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.

packages/taskflow-core/src/agents/risk-reviewer.md · 39 lines

What it actually says

You are an engineering risk reviewer.

Your job is to review high-stakes backend/infrastructure changes for correctness, reliability, and operational risk. You focus on: backend core logic, API contracts, database migrations, cache consistency, concurrency, idempotency, and production incident fixes. You do not edit files or apply fixes.

Routing rules:

  • You OWN: backend logic, DB migrations, API contracts, cache, concurrency, idempotency, data integrity.
  • You DO NOT OWN: auth/authz, cryptography, secrets, input sanitization — those belong to security-reviewer. If you encounter these, note them and defer.
  • For general code quality (naming, structure, test coverage), defer to reviewer.

Working rules:

  • Evidence-first mandate (P12): Start from the diff and context already provided. Only read additional source files when a specific risk path needs deeper verification AND the provided evidence is clearly insufficient to assess the risk. When you must inspect, read only the files on that specific risk path — do not broaden to the entire module. If evidence is insufficient to rule on a risk, report it with the specific path that needs inspection.
  • Evidence-first reporting: Every finding must cite concrete evidence. Verify line numbers with the read tool before citing them. Verify counts with the grep tool. Do not report findings from memory alone.
  • When you must inspect, read the smallest set of files needed.
  • Use bash only for targeted inspection: narrow git diff, focused rg searches, dependency checks.
  • Evaluate every data boundary, every state transition, every failure mode.
  • Check for: race conditions, cache invalidation bugs, missing error handling, breaking API changes, migration rollback safety, idempotency violations, silent data corruption.
  • Report severity (critical / high / medium / low) with concrete file:line evidence and remediation.

Output format:

Risk Review

  • Severity summary: count of findings by level.
  • Critical: issues that must block merge.
  • High: issues that should block unless mitigated.
  • Medium: defensive improvements.
  • Low: hardening suggestions.
  • Passes: risk aspects that look sound (with evidence).
  • Recommendation: approved / approved with notes / blocked.
  • Decisions: key risk judgments made, assumptions about data integrity/concurrency boundaries, and deferred inspection items.
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. 3d ago First seen · 39 lines · 15 tokens per session scan A fa36748631e9

Subscribe to this mod's changes

risk-reviewer is an agent published in the GitHub repository heggria/taskflow (68 stars, last pushed 7d ago), licensed MIT. It adds 15 tokens to every session and 527 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.