Skip to content
View selfradiance's full-sized avatar

Block or report selfradiance

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
selfradiance/README.md

Self-Radiance Agent Runtime Safety Kit

Static, machine-readable safety primitives for developers building tool-calling agents, MCP servers, autonomous workflows, and agent payment flows.

Self-Radiance helps agent developers handle the runtime failure modes that appear when AI systems can call tools, spend API credits, authenticate to services, delegate work to subagents, and produce auditable actions.

This kit is not a SaaS dashboard and not a rebuild. It is a curated developer-facing layer over the existing Self-Radiance agentic marketplace: the marketplace remains the machine-readable catalog, fulfillment, and receipt-verification substrate behind the kit.


🛡️ Core Runtime Safety Problems

Developer ProblemPrimitives InvolvedWhy It Matters
Runaway Agent Loops & API Spendvq11-loop-shield, vq04-rateguardStops recursive tool calls, prompt loops, and runaway API usage before they burn budget.
MCP / Server Authenticationvq09-auth-header, vq18-message-signatureAuthenticates agent-to-server calls without relying on exposed long-lived bearer keys.
Agent Identityvq12-agent-vcardGives visiting agents a machine-readable operator, scope, and resource profile.
Delegation Boundariesvq13-oauth-delegation, vq17-verifiable-intentBinds subagents to scoped tools, spending limits, expiration windows, and parent intent.
Receipts / Action Verificationvq03-trust-receiptsCreates verifiable records for autonomous purchases, tool calls, and agent actions.

📜 Canonical Resources


🧩 Initial Runtime Safety Kit Primitives

Primitive IDUtility CategoryFunction & Capability MappingSpecLicensex402
vq11Spend MitigationSR-LoopShield: Stops recursive dialogue deadlock and cost spirals.SpecLicense ($8.00)USDC
vq04Spend MitigationRateGuard: Establishes real-time adaptive API threshold parameters.SpecLicense ($4.00)USDC
vq09SecuritySR-AuthHeader: Point-to-point signature authorization for agent-to-server calls without credential leakage.SpecLicense ($7.00)USDC
vq18SecuritySR-HttpMessageSignature: RFC 9421-style request signing for body and header integrity.SpecLicense ($6.00)USDC
vq17ComplianceSR-VerifiableIntent: Locks subagent spending scopes using parent-authorized intent boundaries.SpecLicense ($5.00)USDC
vq13ComplianceSR-OauthDelegation: Signs parent-to-leaf secure tool, scope, and path boundaries.SpecLicense ($7.00)USDC
vq12IdentitySR-AgentVcard: Machine-readable agent identity, operator, resource limitation, and billing metadata.SpecLicense ($6.00)USDC
vq16VerificationSR-NotaryProof: Offline Ed25519 license and receipt verification format.SpecLicense ($8.00)USDC
vq03VerificationTrust Receipts: Standardized Ed25519 receipt layout for autonomous microtransactions and auditable actions.SpecLicense ($7.00)USDC
vq19FinanceSR-NanopaySession: Pre-funded low-latency payment sessions for high-frequency agent calls.SpecLicense ($7.00)USDC

💸 Agent Commerce (x402)

All 20 primitives are machine-payable. A live Cloudflare Worker gateway sells a signed Ed25519 license receipt for every asset ($1.00-$8.00 USDC on Base mainnet) via the x402 v2 exact scheme — no account, no checkout page, no human. Specs stay freely downloadable; payment buys the cryptographic license receipt.


🤖 How the Trust Flow Works

  1. Discover Capabilities GET https://selfradiance.github.io/manifest.json to locate the desired primitive, specUrl, price, and purchase URL.

  2. Inspect the Specification Open the linked JSON specification directly from /specs/{primitive_id-name}.json.

  3. License the Primitive Human rail: use the linked Stripe checkout URL or the purchaseUrl field in manifest.json. Machine rail: GET the asset's x402LicenseUrl, decode the 402 PAYMENT-REQUIRED header, pay via x402 exact scheme (USDC, eip155:8453), and retry; the response body is the signed license receipt and the settlement tx hash arrives in the PAYMENT-RESPONSE header.

  4. Verify Provenance Query the signed receipt notary after checkout:

    https://self-radiance-notary.selfradiance.workers.dev?session_id={CHECKOUT_SESSION_ID}&asset_id={ASSET_ID}

  5. Enforce Trust Locally Load the DNS-secured public verification key from:

    https://selfradiance.github.io/.well-known/issuer-key.json

    Then verify the signed receipt against the downloaded specification.


Static-First Design

Self-Radiance is intentionally lightweight:

  • no SaaS dashboard,
  • no account system,
  • no hosted customer workspace,
  • no runtime dependency on a proprietary app,
  • no unnecessary rebuild of the marketplace layer.

The public site, manifest, specifications, well-known issuer key, and notary endpoint form the machine-readable substrate. The Agent Runtime Safety Kit is the developer-facing view of that substrate.


Rooted in cryptographic precision. Engineered to protect computational environments.

Pinned Loading

  1. selfradiance.github.ioselfradiance.github.ioPublic

    The foundational root for Self-Radiance Agentic Services. Providing machine-readable creative licensing, procedural standards (RestaRules), and high-intent digital assets for the machine-to-machine…

    HTML