Agent Tools — text injection-check
Limited track record
How we rate this
Delivery health: unknown. Paid successes: 0. Paid failures: 0. Schema confidence: high.
Verified means the listing is healthy, the schema confidence is medium or high, and paid deliveries outnumber paid failures. Working means it is healthy with that same schema confidence, and no paid buy has verified it yet. Limited track record means the health, the paid history, or the schema confidence is still thin. Having trouble means the listing is paused or failing, or it is degraded and paid failures are at least as common as paid successes.
Detect prompt injection and jailbreak attempts in untrusted text such as scraped web pages, emails and tool output. JSON body {text} (max 50,000 chars).
More
Rule-based detection of instruction override phrasing, role spoofing tags, data leak links and hidden Unicode in 6 languages. Returns verdict safe, suspicious or malicious, a 0-100 score and matched spans. Text is not stored.
Example
What agents use this for. Detect prompt injection and jailbreak attempts in untrusted text such as scraped web pages, emails and tool output. Category: x402.
Request
POST https://agents.crowdspark.me/v1/text/injection-check
curl
curl -sS -X POST \
'https://agents.crowdspark.me/v1/text/injection-check' \
-H 'Content-Type: application/json' \
-d '{}' \
# Example only. No L402 token. Sends no sats.
MCP buy
buy({
"sku": "bazaar-628d87a9ac0a"
})
Response
No response schema is listed, and no example sample is stored.
agents.crowdspark.me