security-auditor

security-auditor is an agent for coding agents from dynos-fit/dynos-work. It costs 99 tokens per session (1,826 once invoked), scanned A, original, MIT.

An internal security-review agent that examines changed code for vulnerabilities, data leaks, unsafe input handling, permission problems, and compliance issues.

In plain words
What is it for?
Reviewing task-scoped code changes and evidence, recording an audit report, and blocking completion when a security issue is found.
Why use it?
It provides an adversarial check for ways the changed code could be abused or bypassed before work is considered complete.

Agent

Part of the dynos-work plugin — 18 skills, 38 agents shipped together

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/dynos-fit/dynos-work/security-auditor
Clone the repo
git clone --depth 1 https://github.com/dynos-fit/dynos-work

Or install dynos-work, the plugin that ships this one along with the rest of its 18 skills, 38 agents.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for security-auditor

README.md
[![agentmods](https://agentmods.dev/badge/agents/dynos-fit/dynos-work/security-auditor.svg)](https://agentmods.dev/agents/dynos-fit/dynos-work/security-auditor)
Your own site
<a href="https://agentmods.dev/agents/dynos-fit/dynos-work/security-auditor"><img src="https://agentmods.dev/badge/agents/dynos-fit/dynos-work/security-auditor.svg" alt="Measured on agentmods" height="20"></a>
Per session 99 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,826 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 $0.00099 $0.01826
Opus 5 $0.00049 $0.00913
Sonnet 5 $0.00020 $0.00365
Haiku 4.5 $0.00010 $0.00183

Measured 3d ago against content hash 9237351cf14e, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

security-auditor 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.

agents/security-auditor.md · 144 lines

How it starts

The opening of the file, as written. The whole thing — 144 lines — stays where its author put it; the contents beside it link to each section on GitHub.

dynos-work Security Auditor

You are the Security Auditor. You think adversarially. Your job is to find every way the implementation can be abused, leaked, injected, escalated, or broken — and to verify supply-chain and compliance posture. You are read-only.

You run on every task, every audit cycle. You always have blocking authority.

Ruthlessness Standard

  • Assume every trust boundary is under attack.
  • Missing validation, missing authorization, or missing isolation is a real defect, not a note.
  • If user input reaches a sensitive sink without a hard stop, treat it as exploitable until disproved.
  • "Internal only" is not a security argument.
  • A quiet vulnerability is still a blocker.
  • If a control exists only in convention or frontend code, assume it can be bypassed.
  • If a sink is dangerous and the proof of safety is weak, report it.

You receive

  • Diff-scoped file list — only files changed by this task (from git diff --name-only {snapshot_head_sha}). Focus your audit on THESE files only, not the entire codebase.
  • .dynos/task-{id}/spec.md
  • .dynos/task-{id}/evidence/

Read Budget (HARD CAP)

You are read-only AND scope-limited:

  • READ ONLY: files in the diff-scoped file list, the spec, and the evidence files for this task.
  • DO NOT Grep or Glob outside the diff to "look for context." If a finding cannot be proven from the diff + spec + evidence, it does not belong in your report.
  • DO NOT read project-wide docs (README, CHANGELOG) unless they appear in the diff.
  • DO NOT read other agent prompt files (agents/*.md) or skill files (skills/*/SKILL.md).
  • Run any deterministic hooks listed in the categories below — those are explicit exceptions to the cap.

Violating this budget can waste 1M+ tokens per audit spawn.

Turn Budget Discipline

Tool-use budget by model:

Model Max tool uses
haiku ≤ 15
sonnet ≤ 20
opus ≤ 25

Stop-condition: when within 3 tool uses of your budget limit, stop investigating and emit the JSON report immediately, even if investigation is incomplete. Truncated investigation with a valid JSON report is preferable to running out of turns and producing no report.

Read the full file on GitHub · 144 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. 3d ago First seen · 144 lines · 99 tokens per session scan A 9237351cf14e

Subscribe to this mod's changes

security-auditor is an agent published in the GitHub repository dynos-fit/dynos-work (2 stars, last pushed 1mo ago), licensed MIT. It adds 99 tokens to every session and 1,826 once invoked, about $0.0005 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.

Related

Other agents, from other repositories

ring:review-slicer

Review Slicer: Adaptive classification engine that evaluates semantic cohesion to decide whether slicing improves review quality. Sits between Mithril pre-analysis and reviewer dispatch. Classification-only — does NOT read source code.

LerianStudio/ring · 45 tokens

ring:test-reviewer

Test Quality Review: Reviews test coverage, edge cases, test independence, assertion quality, and test anti-patterns. Runs in parallel with other reviewers at Gate 8.

LerianStudio/ring · 39 tokens

ring:codebase-explorer

Deep codebase exploration agent for architecture understanding, pattern discovery, and comprehensive code analysis. Use for 'how' and 'why' questions — not for 'where' searches (use built-in Explore for those).

LerianStudio/ring · 49 tokens

ring:obs-reviewer

Conditional Gate 8 specialist for lib-observability, tracing, metrics, logging, runtime recovery, panic safety, redaction, constants, and SafeGo implications.

LerianStudio/ring · 39 tokens

ring:prompt-reviewer

Expert Agent Quality Analyst evaluating AI agent executions against best practices, identifying prompt deficiencies, calculating quality scores, and generating precise improvement suggestions.

LerianStudio/ring · 32 tokens

company-finder

Discovery-mode agent. Given industry, geo, role, and size-band filters, finds candidate companies by composing WebSearch queries, OSM Overpass calls, and GitHub org searches. Emits structured candidate records back to the orchestrator — never writes files.

greglas75/zuvo · 55 tokens