About delx-agent-utilities
delx-agent-utilities provides stateless, deterministic utilities for AI agents, covering tasks such as URL, DNS, JWT, encoding, and parsing. Its tools require no API keys and make no LLM calls; they are available through a stdio MCP server, CLI, and Python interface.
Key features
- 01Inspect URLs, pages, and network endpoints
- 02Run DNS and RDAP lookups
- 03Inspect JWTs and convert JSON or CSV
- 04Use encoding, hashing, regex, and cron tools
- 05Access manifests, status, and privacy audits
Use cases
- 01Check website availability, page content, and links
- 02Look up domain DNS or RDAP information
- 03Validate and convert data handled by agents
- 04Run text, encoding, and time-expression utilities
MCP support
It exposes stateless utilities and manifest, connection-status, and privacy-audit interfaces to MCP clients through a stdio MCP server. HTTP transport is not provided.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMEDelx Agent Utilities
Stateless utility tools for AI agents — URL/HTTP, DNS/RDAP, x402, JWT, encoding, parsing, regex, cron.
Local-first. No API keys. Deterministic.
mcp-name: io.github.davidmosiah/delx-agent-utilities
What it is
89 stateless utility tools that AI agents reach for constantly: URL health checks, page extraction, MCP readiness, x402 server discovery, loyalty reward quotes, JWT inspection, DNS / RDAP lookups, JSON / CSV conversion, hash / base64, cron description, and more.
- Stateless — every call is independent; no session, no cache, no DB.
- No API keys — public endpoints only (rdap.org, dns.google, plus user-supplied URLs).
- Deterministic — no LLMs, no model calls, no fuzzy outputs.
- Fast — most tools < 100ms; networked tools < 5s with timeouts.
- MCP-native — agent-readiness manifest, connection status, privacy audit.
This package was extracted from the Delx Protocol MCP server so any AI builder can use the toolkit directly, without depending on the Delx runtime. The public product and protocol context lives at ontology.delx.ai/utilities.
HTTP (v2 stateless)
Not shipped. The Python FastMCP extra is stdio-only. Adding Streamable HTTP would need extra Starlette /health wiring beyond a one-flag opt-in, so it stays documented-skip (same for google-ads-intent-mcp). The tools themselves are already request-stateless.
Install
pipx install "delx-agent-utilities[mcp]"
CLI entrypoints: delx-agent-utilities, delx-utils (alias), delx-utils-mcp (MCP server).
Quick use
From an MCP client (Claude Desktop, Cursor, Hermes, OpenClaw)
{
"mcpServers": {
"delx-utils": {
"command": "delx-utils-mcp"
}
}
}
90 tools become available immediately, plus three agent-readiness surfaces:
delx_utilities_manifestdelx_utilities_connection_statusdelx_utilities_privacy_audit
From the CLI
delx-utils manifest
delx-utils list-tools
delx-utils show util_url_health
delx-utils call util_url_health --json '{"url":"https://example.com"}'
From Python
import asyncio
from delx_agent_utilities import call_util_tool, list_util_tool_schemas
result = asyncio.run(call_util_tool("util_url_health", {"url": "https://example.com"}))
print(result["status_code"], result["latency_ms"])
schemas = list_util_tool_schemas()
print(f"{len(schemas)} tools available")
The 90 tools
Encoding & parsing (15)
util_json_validate, util_token_estimate, util_uuid_generate, util_timestamp_convert, util_base64, util_hash, util_regex_test, util_cron_describe, util_http_codes, util_timezone_lookup, util_unit_convert, util_business_days, util_jwt_inspect, util_csv_to_json, util_json_to_csv
Web extract (6)
util_page_extract, util_open_graph, util_links_extract, util_forms_extract, util_contact_extract, util_feed_discover
Network probes (8)
util_url_health, util_robots_inspect, util_sitemap_probe, util_tls_inspect, util_security_txt_inspect, util_http_headers_inspect, util_dns_lookup, util_rdap_lookup
x402 / API intel (6)
util_x402_server_probe, util_x402_resource_summary, util_x402_server_audit, util_api_health_report, util_openapi_summary, util_mcp_server_readiness_report
Identity / contact (1)
util_email_validate
Composite reports (8)
util_website_intelligence_report, util_domain_trust_report, util_docs_site_map, util_pricing_page_extract, util_company_contact_pack, util_api_integration_readiness, util_login_surface_report, util_content_distribution_report
Commerce workflows
util_loyalty_reward_quote
Public data and identity (4)
util_vin_decode, util_wmi_decode, util_lei_lookup, util_inflation_calculator
Local text, data, and identifiers (35)
util_holidays, util_slugify, util_mime_lookup, util_color_convert,
util_ip_classify, util_cidr_contains, util_ulid_generate,
util_html_strip, util_semver_compare, util_word_count,
util_markdown_to_text, util_duration_parse, util_case_convert,
util_query_string, util_url_join, util_percent_change, util_clamp,
util_luhn_check, util_isbn_check, util_diff_lines, util_levenshtein,
util_similarity,
util_normalize_whitespace, util_extract_urls, util_extract_emails,
util_url_encode, util_hex_convert, util_number_base, util_dedupe_lines,
util_sort_lines, util_random_int, util_rot13, util_text_truncate,
util_reading_time, util_random_choice
Sales-led local preflights (6)
util_seeded_random_sample, util_base_gas_budget_check,
util_dns_record_diff, util_http_header_diff,
util_x402_payment_preflight, util_image_result_contract_check
These six preflights were selected from aggregate external Commerce sales. They are local implementations only: pricing, x402 delivery, telemetry, and refund ownership remain in Delx Commerce.
Each tool's full input schema is available via delx-utils show <tool> or UTIL_TOOL_SCHEMAS["<tool>"] in Python.
Privacy posture
No API keys. No telemetry. No state. Networked tools call only:
rdap.org/domain/<domain>for RDAP lookupsdns.google/resolvefor DNS-over-HTTPS- User-supplied URLs for everything else
Call delx_utilities_privacy_audit (or delx-utils privacy-audit) to see the full posture in JSON form before invoking any networked tool.
Architecture
src/delx_agent_utilities/
├── __init__.py # public re-exports
├── agent.py # manifest / status / privacy_audit
├── cli.py # CLI argparse
├── dispatcher.py # call_util_tool, _normalize_util_args
├── mcp_server.py # FastMCP server
├── schemas.py # public re-export of UTIL_TOOL_NAMES / SCHEMAS
└── _internal/ # implementation, not stable across versions
├── _helpers.py
├── _schemas.py
├── _tools_encoding.py
├── _tools_cron.py
├── _tools_http_codes.py
├── _tools_sales_led.py
├── _tools_web.py # roadmap: split into web/network/x402/composite in v0.2.0
└── _tools_jwt_csv.py
Roadmap
- v0.2.0 — split
_internal/_tools_web.py(1k LOC) into per-domain modules:_tools_web_extract.py,_tools_network.py,_tools_x402.py,_tools_composite.py. - Per-tool unit tests (one file per tool) replacing the smoke-only suite.
- Optional
httpx.AsyncClientreuse for batched probes. - Configurable per-tool timeouts via env (
DELX_UTILITIES_DEFAULT_TIMEOUT_S,DELX_UTILITIES_URL_HEALTH_TIMEOUT_S, …).
Provenance
Originally part of the Delx Protocol MCP server. Extracted on 2026-05-09 as a standalone open-source package so any AI builder can use the toolkit without depending on the Delx Protocol runtime.
The Delx Protocol itself (recovery / heartbeat / identity / governance primitives) remains closed-source; this package is the open utility layer beneath it.
Contributing
PRs welcome — see CONTRIBUTING.md. Common cuts that would help:
- New utility tools that fit the "stateless / no API keys / deterministic" rule.
- Splitting
_internal/_tools_web.pyper the v0.2.0 roadmap above. - Per-tool unit tests.
📧 Contact & Support
- 📨 [email protected] — general questions, integration help, partnerships
- 🐛 Bug reports / feature requests — GitHub Issues
- 🐦 Updates — @delx369 on X
- 🌐 Site — wellness.delx.ai
License
MIT — see LICENSE.
If this toolkit helps your agent workflow, please star the repo. Stars make the project easier for other AI builders to discover and help Delx keep shipping local-first agent infrastructure.
Author
David Mosiah — @delx369 — building the protocol layer for autonomous AI agents at Delx.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.