Skip to content

Skill Packs

github-actions[bot] edited this page Jun 23, 2026 · 32 revisions

Skill Packs

39 curated multi-tool workflows. Each pack solves a real job that no single tool covers — auditing a domain, working up a time series, decoding an opaque blob, pulling the macro backdrop — and ships as a single MCP prompt. An agent calls prompts/get { name: "<pack>", arguments: { … } } and gets back a ready-to-run plan with the right Agent402 tools wired in (in the right order, with the right inputs).

  • Browse on the live site: agent402.tools/skills (full templates, arguments, examples)
  • MCP discovery: every MCP-aware client picks them up via prompts/listprompts/get
  • Find-by-task: /api/find?q=<task> also recommends a matching pack alongside individual tools, so a task-shaped query points at the workflow, not just the raw tools.

How to call a pack

// MCP (any MCP-aware client — Claude Desktop, Cline, etc.)
prompts/get { "name": "security-audit", "arguments": { "domain": "stripe.com" } }

// HTTP (plain GET — no wallet needed to render a template)
GET https://agent402.tools/skills/security-audit?domain=stripe.com

The template is the plan — the agent then executes each step. Payment (USDC via x402 or free proof-of-work) only happens when the agent actually calls each tool.


Security & trust (4)

Pack What it solves
security-audit Enumerate a domain's external attack surface in one workflow: certs, DNS posture, email auth, HTTP headers, tech stack.
email-deliverability Diagnose why a domain's email lands in spam: SPF posture, DMARC policy, DKIM key strength, MX.
fraud-signals Is this domain trustworthy, or is it a phishing site / typosquat / scam? Pull the reputation signals.
jwt-forensics Someone hands you a JWT and asks "is this valid?" Decode, check expiry, verify the signature.

Web extraction & document intelligence (5)

Pack What it solves
content-extraction Turn arbitrary URLs and PDFs into clean structured text — articles, page metadata, PDF pages.
structured-scrape Pull structured data out of any page deterministically — articles, tables, lists, select-by-CSS.
any-to-markdown "I have a URL but it might be HTML, PDF, or an image — give me clean markdown either way."
document-intel Turn any PDF or image URL into structured data — metadata, text, page ranges, OCR, barcodes.
link-preview Turn a URL into a card-shaped preview — OpenGraph/Twitter metadata + thumbnail.

Finance (5)

Pack What it solves
financial-research SEC filings + real-time quotes + history + macro context for a single ticker.
loan-comparison Compare two or more loan offers on a single rubric (monthly payment, total cost, NPV, IRR).
investment-decision Run a capital allocation decision (equipment, expansion, acquisition) — NPV, IRR, sensitivity.
retirement-planning Will my retirement plan actually work? Project the accumulation phase with compound interest.
savings-goal How much do I need to save each month to hit $X in N years? Pin down the required contribution.

Macro & SEC (4)

Pack What it solves
macro-economics Pull the canonical US macro dataset — yield curve, CPI, unemployment, fed funds, Sahm rule.
macro-context "Is the economic backdrop you're modeling against still current?" — refresh the macro snapshot.
sec-filings-deep-dive Full EDGAR picture of a US public company: filings, facts, insider trades, 13F holdings.
regulatory-watch "Who just filed / who just bought / what just IPO'd / what does the full-text search show?"

Time series & forecasting (2)

Pack What it solves
trend-analysis Take any numeric time series — stock close, FRED indicator, yield — and characterize the trend.
forecasting-bake-off Don't guess which forecasting method to trust. Backtest all four (naive/drift, SES, Holt, Holt-Winters) and pick the winner by MAPE.

Network, DevOps & API work (4)

Pack What it solves
dns-network-ops End-to-end DNS health check: records, multi-resolver propagation, WHOIS, ASN, robots.txt.
status-snapshot "Is this site healthy, addressable, and crawlable — right now?"
api-investigation Point at an unknown API endpoint and figure out how to use it: auth, content type, schema.
schema-evolution "Did this API contract change in a way that breaks our integration?" — diff two OpenAPI specs.

Data engineering & RAG (4)

Pack What it solves
csv-profile Hand it a CSV and get back a column-by-column profile: stats, outliers, correlations.
data-interchange Bring data in from any structured format, normalize through JSON, fan out to YAML/CSV/JSON.
text-hygiene Turn a wall of dirty text into something downstream code can trust — redact, dedupe, sort, extract.
rag-prep Turn a raw document into a vector-DB-ready JSONL dataset, deterministically.

Decoding & inspection (2)

Pack What it solves
decode-blob Hand the agent an opaque string — JWT, base64 JSON, gzipped API response — and identify + decode it.
webhook-debug A webhook hit your endpoint — confirm it's authentic, valid, and safe to log.

Identity & onboarding (2)

Pack What it solves
user-onboarding Take a signup form and run the full onboarding deterministically: validate, hash, slugify.
identity-mint Server-side identity-issuance round-trip: UUIDv4 + deterministic UUIDv5 + slug + password.

Location & time (3)

Pack What it solves
location-intel Point at an address (or rough place name) and assemble the situational brief — geo, weather, alerts.
meeting-scheduler Schedule a meeting across timezones without the round-tripping — convert and present a slot.
trip-planner Plan a multi-stop journey: geocode each stop, sum pairwise distances, add travel + buffer time.

Crypto (1)

Pack What it solves
crypto-research Live price, market structure, OHLC history, trending status, global market context for any coin.

Search & citations (1)

Pack What it solves
search-and-cite Research a question, return an answer with citations. Brave answer + supporting search snippets.

Media & accessibility (2)

Pack What it solves
media-pipeline "User uploaded a thing, normalize it before storing." Probe → resize → thumbnail → convert.
a11y-audit Deterministic WCAG 2.x audit of an HTML page from a string and a fg/bg color pair.

Why packs and not just tools

A single tool answers a question. A pack answers a job.

When an agent says "audit a domain", picking one tool (whois? dns? tls-cert? cert-transparency?) is a guess — the right answer is "all of them in the right order, then synthesize." That's what a pack encodes:

  • The plan is in the template, not in the model. Same pack, same plan, every time — no token-spending discovery loop.
  • The tools are pinned. When a new better tool ships, the pack template gets updated server-side; agents calling prompts/get always get the current best plan.
  • Pricing is transparent. Each tool's price is deterministic; the pack template lists every call so total cost is predictable before the first call.
  • No LLM in the serving path. The pack rendering itself is deterministic — no hidden inference, no surprise dependencies.

Adding a pack

Packs live in src/skills.js. A pack is { slug, title, tagline, useCase, toolSlugs[], arguments[], workflow[], notes[] } — see the existing entries for the shape. CI's "answers its own example" check covers the underlying tools; pack templates are validated by scripts/test-mcp-all.js (prompts/list returns N typed entries; prompts/get renders each one without throwing).

See also

  • Tool Catalog — the underlying 1,280 tools the packs orchestrate
  • MCP Connector — how to wire the connector into Claude / Cline / any MCP-aware client
  • Getting Started — your first call (free, no wallet) in 60 seconds
  • x402-Index-and-Router — what Agent402 looks like inside the wider x402 ecosystem

Clone this wiki locally