Claude-Tools-for-Delphi: Agent for Claude Code

.claude/agents/light-security-ClaudeSettingsAudit.md

light-security-ClaudeSettingsAudit is an agent for Claude Code from GabrielOnDelphi/Claude-Tools-for-Delphi. It costs 85 tokens per session (883 once invoked), scanned A, original, MPL-2.0.

A read-only auditor for Claude Code configuration files such as settings.json, settings.local.json, and .mcp.json. It reports unsafe hooks, credential settings, arbitrary servers, permission bypasses, and overly broad permissions as dangerous, suspicious, or safe.

In plain words
What is it for?
Scanning Claude Code settings, checking them against a known-good baseline, and reporting security findings without editing the files.
Why use it?
It helps find configuration that could run unsafe commands, expose credentials, connect to untrusted services, or grant too much access.

Agent for Claude Code

Written for Claude Code: installed under .claude/. Also seen: model in frontmatter; mentions Claude Code.

This is GabrielOnDelphi/Claude-Tools-for-Delphi's own configuration. It tells Claude Code how to work on Claude-Tools-for-Delphi itself, so it is not a mod to install elsewhere. Copy it as a starting point and replace the rules that are about this project. Everything Claude-Tools-for-Delphi configures →

Reuse

Borrowing it

Nothing to install: this file belongs to GabrielOnDelphi/Claude-Tools-for-Delphi. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file
curl -O https://raw.githubusercontent.com/GabrielOnDelphi/Claude-Tools-for-Delphi/main/.claude/agents/light-security-ClaudeSettingsAudit.md
Clone the repo
git clone --depth 1 https://github.com/GabrielOnDelphi/Claude-Tools-for-Delphi

Made for: Claude Code.

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 light-security-ClaudeSettingsAudit

README.md
[![agentmods](https://agentmods.dev/badge/agents/gabrielondelphi/claude-tools-for-delphi/light-security-claudesettingsaudit/github.svg)](https://agentmods.dev/agents/gabrielondelphi/claude-tools-for-delphi/light-security-claudesettingsaudit)
Your own site
<a href="https://agentmods.dev/agents/gabrielondelphi/claude-tools-for-delphi/light-security-claudesettingsaudit"><img src="https://agentmods.dev/badge/agents/gabrielondelphi/claude-tools-for-delphi/light-security-claudesettingsaudit/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for light-security-ClaudeSettingsAudit

Your own site · 80×15
<a href="https://agentmods.dev/agents/gabrielondelphi/claude-tools-for-delphi/light-security-claudesettingsaudit"><img src="https://agentmods.dev/badge/agents/gabrielondelphi/claude-tools-for-delphi/light-security-claudesettingsaudit.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 85 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 883 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
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 5.1 $0.00085 $0.00883
Opus 5 $0.00043 $0.00441
Sonnet 5 $0.00017 $0.00177
Haiku 4.5 $0.00009 $0.00088

Measured 10d ago against content hash 68ddbb6caaad, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade A, and why

light-security-ClaudeSettingsAudit 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 10d 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/light-security-ClaudeSettingsAudit.md · 47 lines

How it starts

The opening of the file, as written. The whole thing — 47 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You audit Claude Code configuration files for unsafe or malicious content and report every finding classified by severity. You are read-only — you NEVER edit, strip, move, or delete anything. Reporting only.

Step 0 — Load the classification

Read ~\.claude\skills\light-security-ClaudeSettingsAudit\references\settings-danger.md. It is the source of truth: the file types to audit, the DANGEROUS / SUSPICIOUS / SAFE buckets, and the known-good baseline for this machine. If the launching prompt names extra items to trust or extra roots to scan, honor them for this run.

Step 1 — Resolve the scope

Find every Claude Code config file. Glob for these three filename patterns:

  • **/settings.json
  • **/settings.local.json
  • **/.mcp.json

Default coverage is the whole machine, but full-disk globs are slow and some roots are blocked. Scan these roots (each with path: set), in order, and report which you covered:

  1. C:\Users\<you> — user settings + most projects (highest priority)
  2. C:\AI, C:\Projects, C:\Delphi
  3. C:\Program Files\ClaudeCode — managed settings (report if present/unexpected)

Only settings.json / settings.local.json inside a .claude folder (or the managed dir) are Claude config — a stray settings.json from some other app is out of scope; note it and move on. If a root can't be read, say so rather than skipping silently.

Step 2 — Read and classify

Read each file WHOLE (they are small). Do not trust a grep pattern — read the actual command, url, env value, or mcpServers entry and decide. For every field that runs code, sets a credential/proxy env var, or launches an MCP server, classify per the reference:

  • DANGEROUS — hooks running shell/HTTP, apiKeyHelper / aws* / gcp* / otelHeadersHelper / policyHelper / statusLine(command) / fileSuggestion(command), env with *_BASE_URL / *_API_KEY / *_TOKEN / proxy, mcpServers/.mcp.json launching unknown processes or remote URLs.
  • SUSPICIOUS — non-ask permissions.defaultMode, over-broad permissions.allow, enableAllProjectMcpServers, enabledMcpjsonServers, additionalDirectories, disabled protections.
  • SAFE — narrow allow rules, deny rules, non-sensitive env, and anything matching the machine's known-good baseline.

Read the full file on GitHub · 47 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. 10d ago First seen · 47 lines · 85 tokens per session scan A 68ddbb6caaad

Subscribe to this mod's changes

light-security-ClaudeSettingsAudit is an agent published in the GitHub repository GabrielOnDelphi/Claude-Tools-for-Delphi (17 stars, last pushed 2d ago), licensed MPL-2.0. It adds 85 tokens to every session and 883 once invoked, about $0.0004 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.