Prompt inspector

Skill aunicall/prompt-inspector-integration/skills/prompt-inspector

Official SDKs, MCP & Skills for Prompt Inspector. Protect your LLM apps and Agents via Python, Node.js, Skills or Model Context Protocol.

Install

npx -y skills add aunicall/prompt-inspector-integration --skill prompt-inspector

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Detect prompt injection attacks and adversarial inputs in user text before passing it to your LLM. Use when you need to validate or screen user-provided text for jailbreak attempts, instruction overrides, role-play escapes, or other prompt manipulation techniques. Returns a safety verdict, risk score (0–1), and threat categories. Ideal for guarding AI pipelines, chatbots, and any application that feeds user input into a language model.

SKILL.md

3.8 KB, 821 tokens by cl100k_base, as published. Nobody here has run it

Prompt Inspector

Prompt Inspector is a production-grade API service that detects prompt injection attacks, jailbreak attempts, and adversarial manipulations in real time.

📖 For detailed product information, features, and threat categories, see references/product-info.md

Requirements

Provide your API key via either:

Environment variable: PMTINSP_API_KEY=your-api-key, or
~/.openclaw/.env line: PMTINSP_API_KEY=your-api-key

Get your API key at promptinspector.io by creating an app.

Manage custom sensitive words in your dashboard at promptinspector.io.

Commands

Detect a single text (Python)

# Basic detection — prints verdict and score
python3 {baseDir}/scripts/detect.py --text "..."

# JSON output
python3 {baseDir}/scripts/detect.py --text "..." --format json

# Override API key inline
python3 {baseDir}/scripts/detect.py --api-key pi_xxx --text "..."

Detect a single text (Node.js)

# Basic detection
node {baseDir}/scripts/detect.js --text "..."

# JSON output
node {baseDir}/scripts/detect.js --text "..." --format json

# Override API key inline
node {baseDir}/scripts/detect.js --api-key pi_xxx --text "..."

Batch detection from a file (Python)

# Each line in the file is treated as one text to inspect
python3 {baseDir}/scripts/detect.py --file inputs.txt

# JSON output for automation
python3 {baseDir}/scripts/detect.py --file inputs.txt --format json

Output

Default (human-readable)

Request ID : a1b2c3d4-...
Is Safe    : False
Score      : 0.97
Category   : prompt_injection, jailbreak
Latency    : 34 ms

JSON (`--format json`)

{
  "request_id": "a1b2c3d4-...",
  "is_safe": false,
  "score": 0.97,
  "category": ["prompt_injection", "jailbreak"],
  "latency_ms": 34
}

Threat Categories

Prompt Inspector detects 10 threat categories:

instruction_override
asset_extraction
syntax_injection
jailbreak
response_forcing
euphemism_bypass
reconnaissance_probe
parameter_injection
encoded_payload
custom_sensitive_word

📖 For complete category descriptions, see references/product-info.md

API at a Glance

POST /api/v1/detect/sdk
Header: X-App-Key: <your-api-key>
Body:   {"input_text": "<text to inspect>"}

Response:

{
  "request_id": "string",
  "latency_ms": 34,
  "result": {
    "is_safe": false,
    "score": 0.97,
    "category": ["prompt_injection"]
  }
}

Full API reference: docs.promptinspector.io

Notes

Keep text under the limit for your plan tier. Very long inputs may be rejected with HTTP 413.
Use --format json when piping output to other tools.
For bulk workloads, batch requests with --file to minimise round-trip overhead.
Contact [email protected] for enterprise plans and self-hosting support.

Gives 0 of the 12 instructions most prompt engineering skills give in 821 tokens

Counted across 563 of the 626 authors here whose files we hold, read 2026-08-06

ask at most three clarifying questionsin 22 of 563, across 15 files
respond in the user input languagein 14 of 563, across 9 files
preserve the original intentin 13 of 563, across 11 files
Establish baseline metrics and collect representative examplesin 12 of 563, across 2 files
Identify failure modes and prioritize high-impact fixesin 12 of 563, across 2 files
Apply prompt and workflow improvements with measurable goalsin 12 of 563, across 2 files
Roll back quickly if quality or safety metrics regressin 12 of 563, across 2 files
validate changes with tests and roll out in controlled stagesin 12 of 563, across 2 files
generate quantitative baseline performance reportsin 12 of 563, across 2 files
create representative test scenariosin 12 of 563, across 2 files
treat prompts as codein 12 of 563, across 5 files
test prompts on diverse inputsin 12 of 563, across 8 files

Said here and by no other author read

run the python detection script for single text
inspect a file of inputs line by line
use the json format flag when piping output
batch requests for bulk workloads

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.