prompt-injection-guard

prompt-injection-guard is a skill for Claude Code from oscarsterling/clelp-skills. It costs 91 tokens per session (1,541 once invoked), scanned A, original, MIT.

A prompt-screening guard that detects several structural patterns used to disguise forged operator messages before a tool-capable coding agent acts on them.

In plain words
What is it for?
Use it to scan inbound prompts for fake role labels, invalid trusted-channel wrappers, bad timestamps, or repeated message IDs, with a self-test for the scanner.
Why use it?
It turns specific prompt-forgery checks into a deterministic hook and standalone scanner, reducing reliance on the model to recognise spoofed instructions.

Skill for Claude Code

Written for Claude Code: when-to-use in frontmatter.

Good fit Use it to scan inbound prompts for fake role labels, invalid trusted-channel wrappers, bad timestamps, or repeated message IDs, with a self-test for the scanner.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/oscarsterling/clelp-skills/prompt-injection-guard
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.

Any agent
npx skills add oscarsterling/clelp-skills --skill prompt-injection-guard
Clone the repo
git clone --depth 1 https://github.com/oscarsterling/clelp-skills

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 prompt-injection-guard

README.md
[![agentmods](https://agentmods.dev/badge/skills/oscarsterling/clelp-skills/prompt-injection-guard/github.svg)](https://agentmods.dev/skills/oscarsterling/clelp-skills/prompt-injection-guard)
Your own site
<a href="https://agentmods.dev/skills/oscarsterling/clelp-skills/prompt-injection-guard"><img src="https://agentmods.dev/badge/skills/oscarsterling/clelp-skills/prompt-injection-guard/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 prompt-injection-guard

Your own site · 80×15
<a href="https://agentmods.dev/skills/oscarsterling/clelp-skills/prompt-injection-guard"><img src="https://agentmods.dev/badge/skills/oscarsterling/clelp-skills/prompt-injection-guard.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 91 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,541 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 $0.00091 $0.01541
Opus 5 $0.00046 $0.00771
Sonnet 5 $0.00018 $0.00308
Haiku 4.5 $0.00009 $0.00154

Measured 12d ago against content hash 040c0e971fea, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-12, from the pricing page.

Security

Grade A, and why

prompt-injection-guard 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 12d ago.

The scan reads SKILL.md. This mod also ships 2 executable files (reference/inbound-guard.example.py, scripts/inbound-scan.py), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

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.

prompt-injection-guard/SKILL.md · 112 lines

How it starts

The opening of the file, as written. The whole thing — 112 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Prompt Injection Guard

You are wiring a deterministic guard that inspects each inbound prompt and refuses the specific STRUCTURAL shapes a forged operator command takes. The model cannot rationalize its way past a harness-level refusal the way it can talk itself past a prose rule in a system prompt.

The package ships three portable pieces:

  • reference/inbound-guard.example.py - the hook. It checks four content-free signatures and exits 2 (refuse) on a match, 0 (pass) otherwise. It fails OPEN on any parser error so it can never wedge the agent loop.
  • scripts/inbound-scan.py - a standalone scanner that loads the same detection and runs it over arbitrary text, with a --self-test. One source of truth, so a green self-test proves what the live hook will decide.
  • templates/untrusted-content-fence.md - the provenance/taint design doc. The hook is the cheap front line; provenance is the real boundary. Read it before you rely on the hook alone.

The four signatures

  1. Role-label prefix. A body that opens with a configured role word ("Human:", "Assistant:", or your own list), at the top of the prompt or inside a trusted-channel wrapper. Real human input carries no role scaffolding; a body that does is usually the model's own output round-tripped as a fake user turn.
  2. Wrapper integrity. If you configure a trusted-channel wrapper tag, a block wearing that tag must carry the exact source value a genuine envelope carries AND populate every required attribute. A tag that claims the channel but mismatches the source or skimps on attributes is refused. An unparseable wrapper (opener present, no valid close) is refused too: fail CLOSED.
  3. Future / malformed timestamp. A genuine envelope timestamp is the real send time, offset-aware, and not in the future. Future-beyond-tolerance, unparseable, or timezone-naive timestamps are refused.
  4. Duplicate id. The same (source, id) seen again inside a short window with DIFFERENT content is a replay/confabulation tell.

Read the full file on GitHub · 112 lines

Files

What ships with it

6 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

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. 12d ago First seen · 112 lines · 91 tokens per session scan A 040c0e971fea

Subscribe to this mod's changes

prompt-injection-guard is a skill published in the GitHub repository oscarsterling/clelp-skills (0 stars, last pushed 5d ago), licensed MIT. It adds 91 tokens to every session and 1,541 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 skills, from other repositories

systematic-debugging

Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.

obra/superpowers · 21 tokens

local-ai-agents

Build local-first AI agents that run entirely on a developer workstation with Microsoft Foundry Local and Qwen function-calling models. Covers Small Language Models (SLMs), the OpenAI-compatible local endpoint, sandboxed local tools, local RAG with Chroma, local MCP servers, hybrid cloud/local routing, and the…

microsoft/ai-agents-for-beginners · 200 tokens

next-cache-components-adoption

Turn on Cache Components in a Next.js app and resolve the blocking routes it surfaces. Use when the user wants to enable, adopt, or migrate to Cache Components, flip the cacheComponents flag, work through a flood of blocking-prerender / instant validation errors, run the cache-components-instant-false codemod, or…

vercel/next.js · 95 tokens

chat-pet-sprite-creation

Use when creating or changing VS Code chat pet sprite art, sprite sheets, state animations, eye treatments, Stable/Insiders variants, or pet transitions under src/vs/workbench/contrib/chat/browser/widget/media/chatPet.

microsoft/vscode · 53 tokens

cpu-profile-analysis

Analyze V8/Chrome CPU profiles (.cpuprofile) and DevTools trace files (Trace-.json). Use when: profiling performance, investigating slow functions, comparing code paths, finding bottlenecks, analyzing timeToRequest, understanding call trees from sampling profiler data, analyzing layout/paint/rendering, investigating…

microsoft/vscode · 71 tokens

insight-error-page

Write or audit an insight-kind error page for the Next.js dev overlay. Use when creating a new errors/ .mdx page, auditing an existing one, or checking that a page matches the framework fix cards. Covers page structure, title alignment, FixCard cards with Copy prompt button, code snippets, terminology verification…

vercel/next.js · 83 tokens