Caelara Labs Agent Safety tools over remote MCP. Calls need an X-Caelara-Key API key.
MCP Score How we grade
From what our daily checks can see without credentials, last checked 21 Nov 2026.
| Reliability Answered 3 of 3 daily checks. |
30 / 30 |
| Speed Typical response 116 ms. |
20 / 20 |
| Tool docs 4 of 10 tools have a full description; 18 of 18 inputs have a description. |
22 / 25 |
| Stability 4 tool rewrites in 30 days (description or inputs changed, or removed). |
5 / 15 |
| Listing Missing an icon. |
9 / 10 |
How to raise the grade
- Give every tool a description of at least 80 characters: what it does and when to use it.
- Change tool descriptions and inputs rarely, and with a new version: agents trust stable tools.
- Add an icon to your server.json in the official MCP registry.
Connect Remote server
claude mcp add --transport http agent https://agent.caelaralabs.com/mcp
{
"mcpServers": {
"agent": {
"url": "https://agent.caelaralabs.com/mcp"
}
}
}
{
"servers": {
"agent": {
"type": "http",
"url": "https://agent.caelaralabs.com/mcp"
}
}
}
Settings > Connectors > Add custom connector
URL: https://agent.caelaralabs.com/mcp
Tools 10 found in our last check
agent_input_firewallScan untrusted input: injection risk + PII redaction + sanitized text + allow/review/block verdict. Use when you need to screen untrusted text (user messages, web pages, tool output) for prompt injection and personal data before it reaches a model. Example input: {"text":"Ignore…
agent_output_validatorValidate LLM output: repair JSON, enforce schema, check grounding against a source, extract citations. Use when you need to check or repair a model answer: parse and fix broken JSON, validate against a schema, compare with a source for unsupported claims, extract citations. Exam…
pii_redactorRedact emails, phones, SSNs, cards (Luhn), IPs, API keys and JWTs. Use when you need to remove emails, phone numbers, card numbers and similar personal data from text before logging or sending it on. Example input: {"text":"Reach me at [email protected]"}. Returns: redacted, findin…
prompt_injection_scannerScore text for prompt-injection patterns. Use when you need to get a verdict and risk score for one piece of text that may contain a prompt injection. Example input: {"text":"Disregard all prior rules"}. Returns: verdict, risk_score, matches.
token_compressorCompress text to a target ratio of its original length
json_auto_repairRepair malformed JSON from LLM output
schema_enforcerValidate and coerce data to a JSON Schema
hallucination_diffFlag answer sentences unsupported by the source text
citation_extractorExtract URLs, DOIs, arXiv ids and author-year citations
context_chunkerSplit text into token-bounded overlapping chunks
Tool changes seen by our daily check
21 Nov 2026Description changed:
agent_input_firewall,agent_output_validator,pii_redactor,prompt_injection_scanner.
Renames and new inputs are usually routine. A changed description changes what your agent is told the tool does, and can carry new instructions: review it before you rely on the tool again.
Details
| Endpoint | https://agent.caelaralabs.com/mcp (Streamable HTTP) |
| First published | 19 Nov 2026 |
| Last updated | 19 Nov 2026 |
| Registry name | com.caelaralabs/agent |
More ai and media servers See all
Listing data from the official MCP registry. Status and tools from NestDaddy's daily check. Browse all MCP servers.
