agent-manifest.json
A deterministic inventory of pages, resources, tools, policies, approval requirements, and action schemas.
Arachne does not hand you vague AI-readiness advice. It produces a concrete agent interface, evidence, risk controls, runtime governance, monitoring, reports, and the bundle your customer or team needs to connect.
The interface package turns a website into structured tools, resources, schemas, policies, and evidence. It is the customer-facing proof that the site has been compiled into an agent-operable surface.
A deterministic inventory of pages, resources, tools, policies, approval requirements, and action schemas.
A capability inventory where every tool carries source mode, risk level, confidence, approval requirement, source URL, and evidence.
Read, draft, submit, purchase, auth, regulated, and blocked workflows classified for policy enforcement.
A defensible map of where each capability came from: native spec, observed network call, form, extraction path, or manifest.
A summary of read, draft, write, purchase, regulated, and blocked surfaces across the compiled shadow.
Tool and resource definitions that MCP clients can call through the hosted gateway or local config.
The runtime is built for accountability. Reads stay easy. Writes, submissions, purchases, and sensitive workflows route through policy, approval, scoped capability tokens, and audit logs.
Your Shadow API ships as a single Streamable HTTP URL that Claude Desktop, IDE extensions, and agent frameworks can connect to.
State-changing actions mint single-use, merchant- and amount-locked capability tokens that require human approval to activate.
Every capability issued, approval, refusal, and execution is hash-chained into a tamper-evident audit trail.
The operator surface keeps the shadow reviewable after compile. It gives your team the artifacts for onboarding, support, renewal conversations, and site-change monitoring.
Compiled shadows, readiness, compile modes, drift alerts, usage, ownership state, and audit status in one operator view.
Overview, routes, tools, evidence, action registry, risk policies, approvals, audit logs, drift alerts, and settings.
Build, readiness, evidence, drift, and usage-audit reports for email, handoff, customer proof, and internal review.
Rescans compare the latest site against the saved baseline and flag route, tool, schema, selector, confidence, and risk changes.
One download with manifest, MCP config, README, install steps, policy walkthrough, and domain-verification instructions.
A plain-English assessment of discoverability, action risk, forms, auth surfaces, prompt-injection exposure, and remediation steps.
{
"protocol": "artemis-agent-interface",
"site": "https://customer.com",
"verified_domain": "customer.com",
"tools": [
{
"name": "submit_inquiry",
"method": "POST",
"risk": "medium",
"requires_human_approval": true,
"governance": {
"source_mode": "observed",
"confidence": 0.92,
"risk_level": "draft",
"evidence": ["network:POST /api/v1/inquiry"]
}
}
],
"artifacts": [
"normalized-manifest.json",
"normalized-capabilities.json",
"evidence-map.json",
"risk-profile.json"
],
"reports": ["build", "readiness", "evidence", "drift", "usage-audit"],
"hosted": {
"transport": "streamable-http",
"endpoint": "https://mcp.solsticestudio.ai/mcp/shadow_xxx/mcp"
}
}
The free score tells you where the site stands today. The full report and hosted Shadow API turn that scan into a governed agent interface.