# GOSCE Agents

> A portfolio of production remote MCP + A2A agents on getvda.ai, generated by
> GOSCE (GitHub Open-Source Combination Engine). Every agent is capability-scoped,
> speaks open A2A and MCP, carries an Ed25519-signed Agent Card, and meters
> execution over the x402 protocol. Discovery is always free, and an agent
> holding USDC on Base can pay for execution with no account and no signup.

## Start here

- [Portfolio Router](https://router.getvda.ai): describe a capability, get routed
  to the right verified agent. One endpoint instead of 93.
- [Agent index](https://agents.getvda.ai/agents): every agent, its capability and endpoints.
- [Public keys](https://agents.getvda.ai/.well-known/jwks.json): verify any agent card was
  issued by getvda.ai (RFC 7517 JWKS, EdDSA).

## How to call an agent

Each agent exposes, at https://<agent>.getvda.ai:

## Microsoft 365 / Copilot Studio

These are remote streamable-HTTP MCP servers requiring NO authentication, which
is what the Copilot Studio onboarding wizard accepts: Tools > Add a tool > New
tool > Model Context Protocol; Server URL https://router.getvda.ai/mcp;
Authentication None; Create > Create a new connection > Add to agent.
Or import the ready-made connector at <agent>/copilot-studio.yaml, which already
carries x-ms-agentic-protocol: mcp-streamable-1.0 and is ASCII-only so it passes
validation. No store certification is needed to use them.

- `/mcp` - MCP streamable-http. `tools/list`, `server/discover` and the
  `selftest` tool are FREE; other `tools/call` is metered. Protocol 2026-07-28
  (stateless; route on Mcp-Method).

## Paying for a call (agents: no account needed)

Metered calls answer HTTP 402 with a base64 `PAYMENT-REQUIRED` header carrying
an x402 v2 PaymentRequired body. Its `accepts` array offers TWO rails; pick
whichever you can satisfy:

- `exact` on `eip155:8453` (Base mainnet) - the scheme in the x402
  specification. Sign an EIP-3009 `transferWithAuthorization` for the stated
  `amount` in USDC (0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913), return it
  base64 in the `PAYMENT-SIGNATURE` header, and retry. NO account, no signup
  and no prior relationship with us - only a funded wallet. You need no ETH
  either: the facilitator submits the transaction and pays the gas. The
  settlement receipt comes back in the `PAYMENT-RESPONSE` header.
- `nvm:card-delegation` - Stripe card checkout, for humans buying with a card.

Prices are per TOOL and stated in the 402 at request time, in atomic units
(USDC has 6 decimals). Most tools are $0.025. Tools that fan out to hosts you
choose cost more because they cost us more - `health_check` resolves DNS,
completes an SSL handshake per host and fetches up to 50 URLs, and is $0.15.
Read `accepts[].amount`; do not assume a flat rate.

You are never charged for a failed call. Settlement runs only after the agent
returns a successful result, so an authorization for a call that errored is
left unused and expires.

## Verify an agent yourself (free, no key, no signup)

Call the `selftest` tool over MCP, or GET `/selftest`. The agent runs its REAL
capability on a fixed input and returns the genuine result plus a trust manifest
(signed card, JWKS, licence chain, safety posture) with a proof link per claim.

Each result is graded honestly:
- `asserted_correct` - a property of the output was checked.
- `ran_without_error` - it ran and did not raise; nothing was asserted about the
  output. This is NOT a pass and is not presented as one.

## Evidence

Every self-test is sealed to VDA Witness as a tamper-evident, hash-chained,
Ed25519-signed record. /selftest returns the FULL signed record; POST it as
{"record": ...} to https://witness.getvda.ai/api/witness/verify - PUBLIC, no API
key - and you get signatureValid. Alter the record and it returns false.

These are OUR seals about OURSELVES: binding confidence `api_asserted`, custody
`custodial`. A seal proves 'at time T this account asserted X', NOT that X is true.
No later signing or anchoring upgrades that grade.

A record YOU seal of what YOU observed is independent evidence and is the stronger
path. Mint your own credential at https://witness.getvda.ai/api/witness/test-key
(self-serve, no human), or use customer-managed custody so the record is provable
even against VDA. Offline bundle verification: https://witness-dao.getvda.ai
(the DAO-governance evidence layer; it does not accept remote seals).
- `/a2a/` - A2A `message/send`.
- `/.well-known/agent-card.json` - signed Agent Card (also at `agent.json`).
- `/governance.md` - licence chain and data-flow map.
- `/verify` - how to verify the card signature.
- `/selftest` - a real capability result on canned input, free.
- `/llms.txt` - this, per agent.

## Agents

- [Hash passwords with bcrypt and issue/verify JWT session tokens.](https://auth-token-mcp-traced-service.getvda.ai) - `auth`
- [Auth Token MCP Service: composes bcrypt + langchain + mcp + opentelemetry-instrumentation-](https://auth-token-mcp-service.getvda.ai) - `auth`
- [FastAPI Auth Token JWT Service: composes bcrypt + python-jose — crypto, jwt via A2A + MCP.](https://bcrypt-python-jose-1b8dd0.getvda.ai) - `auth`
- [FastAPI Auth Token JWT Service: composes bcrypt + python-jose — crypto, jwt via A2A + MCP.](https://bcrypt-python-jose-d0e0d0.getvda.ai) - `auth`
- [FastAPI Auth Token Service: composes bcrypt + python-jose — crypto, jwt via A2A + MCP.](https://fastapi-auth-token-service.getvda.ai) - `auth`
- [Authenticated LLM JWT Agent: composes bcrypt + openai + python-jose — crypto, llm-client, ](https://authenticated-llm-jwt-agent.getvda.ai) - `auth-llm`
- [Authenticated MCP Agent: composes bcrypt + langchain-core + mcp + opentelemetry-sdk + pyth](https://authenticated-mcp-agent-aed9d5.getvda.ai) - `auth-llm`
- [Authenticated MCP Agent (Bcrypt): composes bcrypt + mcp + openai + opentelemetry-sdk + pyt](https://authenticated-mcp-agent-bcrypt.getvda.ai) - `auth-llm`
- [Authenticated MCP Agent: composes bcrypt + cryptography + langchain-core + mcp + opentelem](https://authenticated-mcp-traced-agent.getvda.ai) - `auth-llm`
- [Authenticated MCP Agent (Bcrypt): composes bcrypt + cryptography + langchain-core + mcp + ](https://bcrypt-cryptography-langchain-cor-027cf2.getvda.ai) - `auth-llm`
- [Authenticated MCP Agent (Bcrypt): composes bcrypt + langchain-core + mcp + openai + opente](https://bcrypt-langchain-core-mcp-25b8f1.getvda.ai) - `auth-llm`
- [Authenticated MCP Agent: composes bcrypt + langchain-core + mcp + openai + opentelemetry-a](https://authenticated-mcp-agent-openai.getvda.ai) - `auth-llm`
- [Authenticated MCP Agent (Bcrypt): composes bcrypt + langchain-core + mcp + opentelemetry-s](https://authenticated-mcp-agent-langchain-core.getvda.ai) - `auth-llm`
- [Authenticated MCP Agent (Bcrypt): composes bcrypt + langchain-core + mcp + openai + opente](https://authenticated-mcp-agent-opentelemetry-api.getvda.ai) - `auth-llm`
- [Authenticated LLM Agent (Bcrypt): composes bcrypt + langchain-openai + mcp + openai + open](https://authenticated-llm-agent-bcrypt.getvda.ai) - `auth-llm`
- [Authenticated LLM Agent: composes bcrypt + langchain-openai + mcp + openai + python-jose —](https://authenticated-llm-agent.getvda.ai) - `auth-llm`
- [Authenticated MCP Agent (Bcrypt): composes bcrypt + mcp + openai + opentelemetry-sdk + pyt](https://authenticated-mcp-agent-bcrypt.getvda.ai) - `auth-llm`
- [MySQL + Redis Compose Generator (Adminer): composes adminer + mysql + redis — tooling, dat](https://mysql-redis-compose-generator-adminer.getvda.ai) - `compose`
- [Memcached + MongoDB + Redis Compose Generator: composes memcached + mongo + redis — cache,](https://memcached-mongodb-redis-compose-generator.getvda.ai) - `compose`
- [MySQL + Redis Compose Generator: composes mysql + redis — database, cache via A2A + MCP.](https://mysql-redis-compose-generator.getvda.ai) - `compose`
- [Postgres + Redis Compose Generator: composes postgres + redis — database, cache via A2A + ](https://postgres-redis-compose-generator.getvda.ai) - `compose`
- [Tells you whether a public MCP server's tools have changed since anyone last looked. We co](https://drift.getvda.ai) - `drift-observatory`
- [Auto-discovering health monitoring. Give it ONE seed — a domain, a URL list, a docker-comp](https://health.getvda.ai) - `health-monitor`
- [Kafka + Postgres + Prometheus Stack Generator: composes confluentinc/cp-kafka + confluenti](https://postgres-prometheus-kafka-stack-generator.getvda.ai) - `infra-stack`
- [Kafka + Postgres Stack Generator: composes confluentinc/cp-kafka + confluentinc/cp-zookeep](https://postgres-kafka-stack-generator.getvda.ai) - `infra-stack`
- [Kafka + MinIO + Redis Stack Generator (Confluentinc/Cp Kafka): composes confluentinc/cp-ka](https://kafka-redis-minio-stack-generator.getvda.ai) - `infra-stack`
- [Kafka + MinIO + Redis Stack Generator: composes confluentinc/cp-kafka + minio/minio + redi](https://kafka-minio-redis-stack-generator.getvda.ai) - `infra-stack`
- [Kafka + Postgres + Prometheus Stack Generator: composes confluentinc/cp-kafka + postgres +](https://postgres-kafka-prometheus-stack-generator.getvda.ai) - `infra-stack`
- [MinIO + Redis Stack Generator: composes confluentinc/cp-zookeeper + minio/minio + redis — ](https://redis-minio-stack-generator.getvda.ai) - `infra-stack`
- [Postgres + Prometheus Stack Generator: composes confluentinc/cp-zookeeper + postgres + pro](https://postgres-prometheus-stack-generator.getvda.ai) - `infra-stack`
- [Elasticsearch + MySQL + Redis Stack Generator: composes elasticsearch/elasticsearch + kiba](https://elasticsearch-redis-mysql-stack-generator.getvda.ai) - `infra-stack`
- [Elasticsearch + Redis Stack Generator: composes elasticsearch/elasticsearch + kibana/kiban](https://elasticsearch-redis-stack-generator.getvda.ai) - `infra-stack`
- [Elasticsearch + MongoDB + Redis Stack Generator (Elasticsearch/Elasticsearch): composes el](https://elasticsearch-mongodb-redis-stack-generator-el-d8f067.getvda.ai) - `infra-stack`
- [Elasticsearch + Redis Stack Generator (Elasticsearch/Elasticsearch): composes elasticsearc](https://elasticsearch-redis-stack-generator-elasticsea-588e3e.getvda.ai) - `infra-stack`
- [Postgres + Grafana Stack Generator: composes grafana/grafana + grafana/loki + postgres — m](https://postgres-grafana-stack-generator.getvda.ai) - `infra-stack`
- [MinIO + Postgres + Redis Stack Generator: composes minio/mc + minio/minio + postgres + red](https://minio-postgres-redis-stack-generator.getvda.ai) - `infra-stack`
- [MinIO + Postgres Stack Generator: composes minio/minio + postgres — storage, database via ](https://postgres-minio-stack-generator.getvda.ai) - `infra-stack`
- [MinIO + Postgres + Redis Stack Generator (Minio/Minio): composes minio/minio + postgres + ](https://redis-minio-postgres-stack-generator.getvda.ai) - `infra-stack`
- [MinIO + Redis Stack Generator: composes minio/minio + redis — storage, cache via A2A + MCP](https://redis-minio-stack-generator-minio-minio.getvda.ai) - `infra-stack`
- [MySQL + Redis + Nginx Stack Generator: composes mysql + nginx + redis — database, web-serv](https://mysql-redis-compose-generator-nginx.getvda.ai) - `infra-stack`
- [Redis + Nginx Stack Generator: composes nginx + redis — web-server, cache via A2A + MCP.](https://nginx-redis-stack-generator.getvda.ai) - `infra-stack`
- [RabbitMQ + Elasticsearch + MongoDB Stack Generator: composes elasticsearch/elasticsearch +](https://rabbitmq-elasticsearch-mongodb-stack-generator.getvda.ai) - `infra-stack`
- [RabbitMQ + Elasticsearch + MySQL Stack Generator: composes elasticsearch/elasticsearch + m](https://rabbitmq-elasticsearch-mysql-stack-generator.getvda.ai) - `infra-stack`
- [RabbitMQ + Elasticsearch + Postgres Stack Generator (Elasticsearch/Elasticsearch): compose](https://rabbitmq-elasticsearch-postgres-stack-generato-053bcf.getvda.ai) - `infra-stack`
- [RabbitMQ + Elasticsearch + Postgres Stack Generator (Mongo): composes elasticsearch/elasti](https://rabbitmq-elasticsearch-postgres-stack-generator-mongo.getvda.ai) - `infra-stack`
- [RabbitMQ + Postgres + MongoDB Stack Generator: composes mongo + postgres + rabbitmq + redi](https://rabbitmq-postgres-mongodb-stack-generator.getvda.ai) - `infra-stack`
- [RabbitMQ + Redis Stack Generator: composes rabbitmq + redis — messaging, cache via A2A + M](https://rabbitmq-redis-stack-generator.getvda.ai) - `infra-stack`
- [GOSCE Portfolio Router — the single entry point that routes orchestrators to verified GOSC](https://router.getvda.ai) - `integration`
- [LLM Orchestration Agent (Python Jose): composes cryptography + langchain-core + mcp + open](https://cryptography-langchain-core-mcp-07d652.getvda.ai) - `langchain-llm`
- [LLM Orchestration Traced Agent: composes cryptography + langchain-core + mcp + opentelemet](https://llm-orchestration-traced-agent.getvda.ai) - `langchain-llm`
- [LLM Orchestration Agent (Cryptography): composes cryptography + langchain-core + mcp + ope](https://llm-orchestration-agent-cryptography.getvda.ai) - `langchain-llm`
- [LLM Orchestration MCP Agent: composes cryptography + langchain + langchain-core + mcp + op](https://llm-orchestration-mcp-agent.getvda.ai) - `langchain-llm`
- [LLM Orchestration Agent (Mcp): composes cryptography + langchain-openai + mcp + openai + o](https://llm-orchestration-agent-mcp.getvda.ai) - `langchain-llm`
- [LLM Orchestration Agent (Langchain Anthropic): composes langchain-anthropic + langchain-co](https://llm-orchestration-agent-langchain-anthropic.getvda.ai) - `langchain-llm`
- [LLM Orchestration Agent (Openai): composes langchain-anthropic + langchain-openai + openai](https://llm-orchestration-agent-openai.getvda.ai) - `langchain-llm`
- [LLM Orchestration Agent 2: composes langchain-core + langchain-openai + langsmith + openai](https://langchain-core-langchain-openai-l-736876.getvda.ai) - `langchain-llm`
- [LLM Orchestration Agent 2: composes langchain-core + langchain-openai + langsmith + openai](https://llm-orchestration-agent-d2feb0.getvda.ai) - `langchain-llm`
- [LLM Orchestration Agent 2: composes langchain-core + langchain-openai + mcp + openai — orc](https://llm-orchestration-agent-2.getvda.ai) - `langchain-llm`
- [LLM Orchestration Agent (Langgraph): composes langchain-core + langgraph + mcp + openai — ](https://llm-orchestration-agent-langgraph.getvda.ai) - `langchain-llm`
- [LLM Observability & Orchestration Agent: composes langchain-core + langsmith + openai — or](https://llm-observability-orchestration.getvda.ai) - `langchain-llm`
- [LLM Orchestration Agent (Langchain Core): composes langchain-core + mcp + openai + opentel](https://llm-orchestration-agent-langchain-core.getvda.ai) - `langchain-llm`
- [LLM Orchestration Observability Agent: composes langchain + langchain-community + langchai](https://llm-orchestration-observability-agent.getvda.ai) - `langchain-llm`
- [LLM Observability & Orchestration Agent (Langchain): composes langchain + langchain-commun](https://llm-observability-orchestration-agent-langchai-c1a755.getvda.ai) - `langchain-llm`
- [LLM Orchestration Agent (Langsmith): composes langchain + langchain-core + langchain-opena](https://llm-orchestration-agent-langsmith.getvda.ai) - `langchain-llm`
- [LLM Observability & Orchestration Agent (Langchain): composes langchain + langchain-core +](https://langchain-langchain-core-langsmit-d042fc.getvda.ai) - `langchain-llm`
- [LLM Orchestration Agent (Langchain Openai): composes langchain + langchain-openai + langgr](https://llm-orchestration-agent-langchain-openai.getvda.ai) - `langchain-llm`
- [LLM Orchestration Agent (Langchain): composes langchain + langchain-openai + langgraph + m](https://llm-orchestration-agent-langchain.getvda.ai) - `langchain-llm`
- [LLM Orchestration Agent: composes langchain-mcp-adapters + langchain-openai + mcp + openai](https://llm-orchestration-agent-d0a1a1.getvda.ai) - `langchain-llm`
- [LLM Orchestration Agent (Opentelemetry Api): composes cryptography + langchain-openai + mc](https://llm-orchestration-agent-opentelemetry-api.getvda.ai) - `langchain-llm`
- [LLM Orchestration JWT Agent: composes cryptography + langchain-core + mcp + opentelemetry-](https://llm-orchestration-jwt-agent.getvda.ai) - `langchain-llm`
- [Observable Backend Stack Generator: composes grafana/grafana + postgres + prom/prometheus ](https://observable-backend-stack-generator.getvda.ai) - `monitoring`
- [Authenticated Multi-LLM Agent: composes anthropic + google-auth-oauthlib + mcp + openai — ](https://authenticated-multi-llm-agent.getvda.ai) - `oauth-llm`
- [Authenticated LLM OAuth Agent: composes anthropic + google-auth-oauthlib + mcp — llm-clien](https://authenticated-llm-agent-anthropic.getvda.ai) - `oauth-llm`
- [Authenticated LLM OAuth Agent: composes cryptography + google-auth-oauthlib + langchain + ](https://authenticated-llm-agent-cryptography.getvda.ai) - `oauth-llm`
- [Authenticated LLM Traced Agent: composes cryptography + google-auth-oauthlib + langchain-o](https://authenticated-llm-traced-agent.getvda.ai) - `oauth-llm`
- [Authenticated LLM OAuth Agent: composes cryptography + google-auth-oauthlib — crypto, oaut](https://authenticated-llm-oauth-agent.getvda.ai) - `oauth-llm`
- [Authenticated LLM OAuth Agent: composes cryptography + google-auth-oauthlib + langchain + ](https://cryptography-google-auth-oauthlib-931267.getvda.ai) - `oauth-llm`
- [Authenticated LLM OAuth Agent: composes google-auth-oauthlib + langchain + langchain-commu](https://google-auth-oauthlib-langchain-la-c1a755.getvda.ai) - `oauth-llm`
- [Authenticated LLM MCP Agent: composes google-auth-oauthlib + langchain + langchain-communi](https://authenticated-llm-mcp-agent.getvda.ai) - `oauth-llm`
- [Authenticated LLM OAuth Agent: composes google-auth-oauthlib + langchain + langchain-commu](https://google-auth-oauthlib-langchain-la-d46b90.getvda.ai) - `oauth-llm`
- [Authenticated LLM OAuth Agent: composes google-auth-oauthlib + langchain + mcp + openai — ](https://authenticated-llm-agent-google-auth-oauthlib.getvda.ai) - `oauth-llm`
- [Structured Output MCP Agent: composes fastmcp + instructor — agent-protocol, structured-ou](https://structured-output-agent.getvda.ai) - `structured-output`
- [Structured Output Traced Agent: composes groq + instructor + langchain-community + opentel](https://structured-output-agent-c2ddcc.getvda.ai) - `structured-output`
- [Structured Output Agent (Litellm): composes instructor + litellm — structured-output, llm-](https://instructor-litellm-6af681.getvda.ai) - `structured-output`
- [Structured Output Agent: composes anthropic + instructor — llm-client, structured-output v](https://structured-output-agent-89bcb0.getvda.ai) - `structured-output`
- [Structured Output Agent (Openai): composes instructor + openai — structured-output, llm-cl](https://structured-output-agent-openai.getvda.ai) - `structured-output`
- [Structured Output MCP Traced Agent: composes instructor + langchain-core + mcp + openai + ](https://structured-output-mcp-agent-instructor.getvda.ai) - `structured-output`
- [Structured Output MCP Traced Agent: composes instructor + langchain-core + mcp + openai + ](https://structured-output-mcp-agent-langchain-core.getvda.ai) - `structured-output`
- [Structured Output MCP Traced Agent: composes instructor + langchain-core + mcp + openai + ](https://structured-output-mcp-traced-agent.getvda.ai) - `structured-output`
- [Traced LLM MCP Proxy: composes anthropic + lm-format-enforcer + mcp + opentelemetry-api + ](https://traced-llm-proxy-anthropic.getvda.ai) - `traced-gemini`
- [Traced LLM Proxy: composes anthropic + mcp + opentelemetry-api + opentelemetry-sdk — llm-c](https://traced-llm-proxy.getvda.ai) - `traced-gemini`
- [Proxy Gemini (Vertex AI) completions wrapped in OpenTelemetry trace spans; returns the ans](https://traced-llm-mcp-proxy.getvda.ai) - `traced-gemini`
- [Proxy Gemini (Vertex AI) completions wrapped in OpenTelemetry trace spans; returns the ans](https://traced-multi-llm-proxy.getvda.ai) - `traced-gemini`

## Notes

- Execution is paid; discovery, governance and self-test are not.
- An unsigned Agent Card means the agent predates signing, not that it is forged.