> ## Documentation Index
> Fetch the complete documentation index at: https://docs.agenticenv.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Configuration

> Configure Go, LLM credentials, Docker dependencies, and env vars before running any example

Set this up once before running any example. Individual example pages only list services or variables needed beyond this base.

## Requirements

| Requirement          | Notes                                                                              |
| -------------------- | ---------------------------------------------------------------------------------- |
| **Go 1.26+**         | `go version` to verify                                                             |
| **LLM API key**      | Set `LLM_APIKEY` in `examples/.env`                                                |
| **Repository clone** | `git clone https://github.com/agenticenv/agent-sdk-go && cd agent-sdk-go/examples` |

Supported providers: `openai`, `anthropic`, `gemini` — configured via `LLM_PROVIDER` and `LLM_MODEL`.

## Environment file setup

Examples load **`examples/.env.defaults`** automatically on startup — do not copy it. Create **`examples/.env`** (gitignored) only for your API key and overrides:

```bash theme={null}
cd examples
cat >> .env <<'EOF'
LLM_APIKEY=your-key-here
LLM_PROVIDER=openai
LLM_MODEL=gpt-4o
EOF
```

Load order: `.env.defaults` → `.env` → process environment. Process-level exports win last. Never commit API keys.

## Optional dependencies

| Dependency                       | When needed                                                                       |
| -------------------------------- | --------------------------------------------------------------------------------- |
| **Docker + Compose**             | Weaviate, pgvector, Redis, Temporal, OTLP collector                               |
| **[Task](https://taskfile.dev)** | `task infra:*:up` targets from `examples/`                                        |
| **Node.js**                      | MCP stdio server (`npx`), AG-UI Next.js UI                                        |
| **Temporal server**              | When `AGENT_RUNTIME=temporal`                                                     |
| **Embedding API key**            | pgvector and some memory/retriever examples — `EMBEDDING_OPENAI_APIKEY` in `.env` |

## Runtime modes

| Mode                | Setting                        | Requirement                                                                    |
| ------------------- | ------------------------------ | ------------------------------------------------------------------------------ |
| **Local** (default) | `AGENT_RUNTIME=local` or unset | None                                                                           |
| **Temporal**        | `AGENT_RUNTIME=temporal`       | Running Temporal server — `task infra:temporal:up && task infra:temporal:wait` |

## Environment variables

### Core

| Variable         | Description                                                    |
| ---------------- | -------------------------------------------------------------- |
| `LLM_PROVIDER`   | `openai`, `anthropic`, or `gemini`                             |
| `LLM_APIKEY`     | API key — **required**                                         |
| `LLM_MODEL`      | e.g. `gpt-4o`, `claude-sonnet-4-20250514`, `gemini-2.5-flash`  |
| `LLM_BASEURL`    | Optional custom or proxy endpoint                              |
| `AGENT_RUNTIME`  | `local` (default) or `temporal`                                |
| `LOG_LEVEL`      | `error` (default), `warn`, `info`, `debug` — written to stderr |
| `SHOW_LLM_USAGE` | `true` to print token usage footer after each run              |
| `SHOW_TELEMETRY` | `true` to print run telemetry footer                           |

### Temporal

Used when `AGENT_RUNTIME=temporal`:

| Variable             | Default                                        |
| -------------------- | ---------------------------------------------- |
| `TEMPORAL_HOST`      | `127.0.0.1`                                    |
| `TEMPORAL_PORT`      | `7233`                                         |
| `TEMPORAL_NAMESPACE` | `default`                                      |
| `TEMPORAL_TASKQUEUE` | `agent-sdk-go` (each example appends a suffix) |

### Conversation

| Variable          | Description                                                        |
| ----------------- | ------------------------------------------------------------------ |
| `REDIS_ADDR`      | Redis address for conversation examples (default `localhost:6379`) |
| `CONVERSATION_ID` | Optional session ID override                                       |

### MCP

| Variable                                                 | Description                                                  |
| -------------------------------------------------------- | ------------------------------------------------------------ |
| `MCP_TRANSPORT`                                          | `stdio` or `streamable_http`                                 |
| `MCP_STDIO_COMMAND` / `MCP_STDIO_ARGS`                   | Local subprocess MCP (default: filesystem sandbox via `npx`) |
| `MCP_STREAMABLE_HTTP_URL`                                | Remote MCP URL                                               |
| `MCP_BEARER_TOKEN`, `MCP_ALLOW_TOOLS`, `MCP_BLOCK_TOOLS` | Auth and tool filters                                        |
| `MCP_CLIENT_ID`, `MCP_CLIENT_SECRET`, `MCP_TOKEN_URL`    | OAuth client credentials                                     |
| `MCP_SKIP_TLS_VERIFY`                                    | Dev only                                                     |

### A2A

| Variable                                                           | Description                             |
| ------------------------------------------------------------------ | --------------------------------------- |
| `A2A_URL`                                                          | Remote agent base URL (client examples) |
| `A2A_TOKEN`, `A2A_HEADERS`, `A2A_ALLOW_SKILLS`, `A2A_BLOCK_SKILLS` | Client options                          |
| `A2A_SERVER_HOST`, `A2A_SERVER_PORT`, `A2A_SERVER_BEARER_TOKENS`   | Inbound server example                  |

### Observability

| Variable                      | Description                                   |
| ----------------------------- | --------------------------------------------- |
| `OTEL_EXPORTER_OTLP_ENDPOINT` | Collector `host:port` — e.g. `localhost:4317` |
| `OTLP_PROTOCOL`               | `grpc` (default) or `http`                    |
| `OTLP_INSECURE`               | `true` for plaintext local collectors         |

### Retriever and memory

| Variable                                                 | Description                        |
| -------------------------------------------------------- | ---------------------------------- |
| `RETRIEVER_MODE`                                         | `agentic`, `prefetch`, or `hybrid` |
| `WEAVIATE_*`                                             | Weaviate host, class, field names  |
| `PGVECTOR_DSN`, `PGVECTOR_TABLE`, `EMBEDDING_*`          | Postgres + embeddings              |
| `MEMORY_USER_ID`, `MEMORY_STORE_MODE`, `MEMORY_RECALL_*` | Long-term memory examples          |

The authoritative default values for all variables live in `examples/.env.defaults` in the repository.

<Card title="Running Examples" icon="play" href="/examples/running-examples">
  Commands, Task infra, suggested order, and multi-process examples
</Card>