secure-llm-features SKILL.md skillB
secure-llm-features from velimirmueller/lab-claude-skills is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.
mdr add velimirmueller/lab-claude-skills/secure-llm-features@git:20261009.b6d037aPins git:20261009.b6d037a in mdr.lock. Pin the fingerprint instead.
- Audit grade
- B, 16 of 17 checks passed
- Kind
- SKILL.md skill
- Repository
- velimirmueller/lab-claude-skills
- Path
skills/ai/secure-llm-features/SKILL.md- Versions recorded
- 1
- Latest version
- git:20261009.b6d037a, committed Oct 9, 2026
- Size
- 7,272 bytes
- Repository stars
- 5
- License
- MIT
Use when shipping or reviewing an LLM feature that reads untrusted text or has tools — prompt-injection threat model, least-privilege tools scoped by the user's token, safe output handling, per-user rate and cost limits, logs without prompt PII; mapped to the OWASP LLM Top 10 (2026).
Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.
What the file says
# Secure LLM Features A model cannot tell instructions from data, so any text it reads (a document, a web page, a ticket, a tool result) can try to steer it. You cannot prompt that away. You design so that a steered model has no power: its authority is the user's token, its writes need a human, its output is escaped like any untrusted input, and its spend has a ceiling. The contract and the OWASP map are in [llm-security.md](../_shared/llm-security.md). ## 1. Audit current state Read `.claude/stack-profile.md` (`languages`, `backend.track`, `frontend`, `ai`). If absent, detect per [stack-profile.md](../../core/_shared/stack-profile.md). Change nothing yet; build the picture: ```bash grep -rnE "llm\.(generate|stream)|messages\.(create|stream|parse)|@mcp\.tool|registerTool" src # every model call and tool grep -rnE "(userId|user_id|tenantId|tenant_id|accountId)" src | grep -iE "tool|schema|z\.object" # identity as a tool input grep -rnE "dangerouslySetInnerHTML|v-html|innerHTML|marked\(|markdown-it|react-markdown" src # output sinks grep -rniE "(logger|console)\.[a-z]+\(.*(prompt|messages|completion|content)" src # prompt content in logs …
Read the whole file at its exact version.
How to install
mdr add velimirmueller/lab-claude-skills/secure-llm-features@sha256:513e53507a2299b9The line at the top of this page pins the label git:20261009.b6d037a, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.
[](https://markdownregistry.com/a/art_uywkrjjsungggwf7)
3 badge image fetches in 30 days. What the badge shows.
Versions
The registry has recorded 1 version of this file, committed on Oct 9, 2026.
Audit of the latest version
- fail: No script tag (matched: <script)
- pass: Frontmatter block present
- pass: Frontmatter declares a name
- pass: Frontmatter declares a description
- pass: Size between 200 bytes and 200 KB (7272 bytes)
- pass: No zero-width or bidi control characters
- pass: No instruction hidden inside an HTML comment
- pass: No link to an exfiltration or paste host
- pass: No credential-shaped string
- pass: No instruction to send local credentials anywhere
- pass: No text hidden with inline styles
- pass: No prompt-injection phrasing
- pass: No curl or wget piped into a shell
- pass: No recursive delete of root, home or parent
- pass: No instruction to read or print local credentials
- pass: No base64 blob over 200 characters
- pass: No link to a raw IP address
Starting a new one? Use a template for SKILL.md on mdpantry, which passes every one of these checks and says what to change in it.
Source
velimirmueller/lab-claude-skills · 5 stars · license MIT · pushed 2026-10-12 · branch main
API
GET https://markdownregistry.com/api/v1/artifacts/art_uywkrjjsungggwf7 GET https://markdownregistry.com/api/v1/resolve?ref=velimirmueller/lab-claude-skills/secure-llm-features GET https://markdownregistry.com/api/v1/blob/513e53507a2299b928377d57d6077ca19757b5a75678956c9fcc3b9227a4899f
Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.
Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.
More from velimirmueller/lab-claude-skills
Every file in velimirmueller/lab-claude-skills