Skip to main content

Infrastructure

Agent Discovery

Sigil exposes structured data at every layer so AI agents can discover capabilities, consume scan results, and interact with tools programmatically. This page covers all the discovery mechanisms.

Discovery stack

Four complementary standards give agents everything they need, from a quick product summary to executable tool schemas:

A2A

Agent Card

Machine-readable “business card” at /.well-known/agent-card.json. Lists Sigil's identity, 7 MCP skills with full input schemas, auth requirements, and API endpoint links. Follows the Google A2A protocol (IANA-registered well-known URI).

WebMCP

Browser-Exposed Tools

Forms and interactive components expose tools via Chrome's WebMCP standard. Newsletter signup, waitlist, and scan search are all registered as callable tools for compatible AI agents.

LLM

llms.txt + llms-full.txt

Human- and LLM-readable plain text files with product summary, install commands, scan phases, pricing, and all documentation links. llms.txt is the concise version; llms-full.txt is the complete technical reference.

SEO

JSON-LD Structured Data

Every page embeds schema.org JSON-LD: SoftwareApplication on the homepage, TechArticle + FAQPage on scan reports, BreadcrumbList on all docs pages, and DigitalDocument on signed attestations.

A2A Agent Card

The agent card tells other agents what Sigil can do. It includes 7 skills (6 MCP tools + attestation verification), each with full JSON Schema for inputs.

bash
curl https://sigilsec.ai/.well-known/agent-card.json | jq '.skills[].id'
Skills
"sigil_scan"
"sigil_scan_package"
"sigil_clone"
"sigil_quarantine"
"sigil_approve"
"sigil_reject"
"sigil_verify_attestation"

The card also includes links to all API endpoints, documentation, and discovery files in the _links section.

FieldValue
schemaVersion1.0
nameSigil
providerNOMARK Pty Ltd
authSchemesnone (CLI runs locally)
skills7 tools with input schemas

WebMCP tools

Sigil uses both declarative HTML attributes and imperative JavaScript registration to expose tools to WebMCP-compatible agents.

Declarative tools (HTML attributes)

Forms annotated with toolname, tooldescription, and toolparamdescription attributes. Ignored by browsers that don't support WebMCP.

ToolComponentParameters
subscribeToNewsletterNewsletter formemail
requestEarlyAccessWaitlist formemail, use_case, team_size

Imperative tools (JavaScript)

The scan database page registers a searchSigilScans tool via navigator.modelContext.registerTool() that lets agents search and filter scan results by ecosystem, verdict, and package name.

Progressive enhancement
WebMCP is Chrome 146+ only (early preview). The attributes and JavaScript are silently ignored by unsupported browsers. No fallback needed — the underlying forms and search work normally for everyone.

llms.txt

Two plain-text files designed for LLM consumption:

llms.txt

Concise product summary, install commands, doc links, scan phases, pricing, attestation endpoints. Under 100 lines.

llms-full.txt

Complete technical reference: CLI commands, API endpoints, MCP tool schemas, Claude Code plugin, CI/CD config, attestation verification, troubleshooting FAQ.

JSON-LD structured data

Schema.org JSON-LD is embedded on every page for search engines and AI agents:

PageSchemaPurpose
HomepageSoftwareApplicationProduct identity, version, pricing, features
Scan reportsTechArticle + FAQPageScan results, “Is X safe?” answers
Signed scansDigitalDocumentAttestation URL, content digest, encoding format
Docs pagesBreadcrumbListNavigation hierarchy
Blog postsArticle + BreadcrumbListAuthor, date, publisher, navigation
Docs with FAQFAQPage + HowToTroubleshooting, quickstart steps

API endpoints for agents

All endpoints return JSON with CORS headers. No authentication required for public data.

EndpointDescription
GET /api/v1/feedScan feed with ecosystem/verdict/since filters
GET /api/v1/attestation/{id}DSSE envelope for a signed scan
GET /api/v1/verifyServer-side attestation verification
GET /badge/{eco}/{pkg}SVG scan badge for README embeds
GET /feed.xmlRSS 2.0 feed of recent scans

Discovery file index

All machine-readable discovery files at a glance:

FileStandardPurpose
/.well-known/agent-card.jsonA2AAgent capabilities, skills, auth
/.well-known/sigil-verify.jsonCustomSigning keys, verification instructions
/llms.txtllms.txtConcise product summary for LLMs
/llms-full.txtllms.txtComplete technical reference
/robots.txtrobots.txtCrawl directives + agent card reference
/sitemap.xmlSitemapAll crawlable URLs including scan pages

See also

  • MCP Server — run Sigil as a tool server for AI agents
  • Scan Attestations — cryptographic verification of scan results
  • Sigil Bot — the autonomous scanner that powers the scan database

Need help?

Ask a question in GitHub Discussions or check the troubleshooting guide.