Using AI Coding Agents with Volary
Volary adds persistent memory to your coding tools. Your agent learns your codebase, your conventions, and your preferences – then applies that knowledge automatically in every session.
Supported tools
Claude Code
Anthropic's CLI coding agent. Proxy through Volary with two environment variables and add the Volary MCP server with one command.
OpenAI Codex
OpenAI's coding agent. Configure Volary as an MCP server and wire up session hooks for automatic transcript capture.
Cursor
AI-powered code editor. Connect to Volary with the Cursor plugin and a local config file.
Kimi
Moonshot's CLI coding agent. Add Volary as a provider in your Kimi config and register the MCP server.
GitHub Copilot
Install the Volary VS Code extension, or register Volary as an MCP server with the Copilot CLI plus session hooks.
Gemini
Google's CLI coding agent. Install the Volary Gemini extension and configure it with your agent details.
API Provider
Any other tool that speaks the OpenAI or Anthropic API, or that supports MCP. Point it at the Volary base URL with your access token.
Getting started
Before configuring your coding tool, set up Volary:
- 1Sign up or log in to the Volary dashboard.
- 2Create an organisation and an agent. The setup wizard will guide you through choosing your tool and configuring an upstream endpoint.
- 3Copy the configuration shown in the setup wizard, or follow the tool-specific instructions below.
Claude Code
Paste the exports below into your shell, then run claude. Replace the placeholder values with your organisation ID, agent ID, and API token from the dashboard.
export ANTHROPIC_BASE_URL="https://api.volary.ai/v0/orgs/{ORG_ID}/agents/{AGENT_ID}"
export ANTHROPIC_CUSTOM_HEADERS="X-Volary-Api-Key: {YOUR_TOKEN}"Your Anthropic API key stays local – Volary uses the upstream endpoint you configured in the dashboard.
MCP Server
Also add the Volary MCP server to Claude Code so it can recall and store memories explicitly:
claude mcp add volary --transport http --header "Authorization: Bearer {YOUR_TOKEN}" https://api.volary.ai/v0/orgs/{ORG_ID}/agents/{AGENT_ID}/v0/mcpOr add it manually to ~/.claude/settings.json (or a project-local .claude/settings.json):
{
"mcpServers": {
"volary": {
"type": "http",
"url": "https://api.volary.ai/v0/orgs/{ORG_ID}/agents/{AGENT_ID}/v0/mcp",
"headers": {
"Authorization": "Bearer {YOUR_TOKEN}"
}
}
}
}OpenAI Codex
Note: hooks are currently an experimental feature of Codex and aren't supported on all platforms. You will need jq installed for the transcript hook to work.
Add the following to your Codex config (usually ~/.codex/config.toml). This registers Volary as an MCP server and exposes your token via the Codex shell environment policy.
[features]
codex_hooks = true
[shell_environment_policy]
include_only = ["PATH", "HOME", "VOLARY_TOKEN"]
ignore_default_excludes = true
[shell_environment_policy.set]
VOLARY_TOKEN = "{YOUR_TOKEN}"
[mcp_servers.volary]
url = "https://api.volary.ai/v0/orgs/{ORG_ID}/agents/{AGENT_ID}/v0/mcp"
http_headers = { "Authorization" = "Bearer {YOUR_TOKEN}" }
Session hooks
Add the following to ~/.codex/hooks.json so Codex retrieves the root memory index at session start and uploads the transcript on stop:
{
"hooks": {
"SessionStart": [
{
"hooks": [
{
"type": "command",
"command": "curl -f -X POST -H \"Authorization: Bearer ${VOLARY_TOKEN}\" -d '{}' https://api.volary.ai/v0/orgs/{ORG_ID}/agents/{AGENT_ID}/copilot/session-start",
"statusMessage": "Retrieving root memory index...",
"timeout": 30
}
]
}
],
"Stop": [
{
"hooks": [
{
"type": "command",
"command": "curl -fsS -X POST -H \"Authorization: Bearer ${VOLARY_TOKEN}\" --data-binary @\"$(cat | jq -r .transcript_path)\" https://api.volary.ai/v0/orgs/{ORG_ID}/agents/{AGENT_ID}/transcripts/codex && echo '{}'",
"statusMessage": "Transcribing memories...",
"timeout": 30
}
]
}
]
}
}
Codex doesn't always inject environment variables into hooks reliably. If hooks fail with a missing VOLARY_TOKEN, export it in your shell before launching Codex.
Cursor
Install the Volary Cursor plugin:
mkdir -p ~/.cursor/plugins/local
git clone git@github.com:volary-ai/cursor-plugin.git ~/.cursor/plugins/local/volary
Then create ~/.cursor/volary.json with your details from the dashboard:
{
"org_id": "{ORG_ID}",
"agent_id": "{AGENT_ID}",
"token": "{YOUR_TOKEN}"
}Kimi
Add Volary as a provider in your Kimi config (usually ~/.kimi/config.toml):
# Set volary as the default model (optional - update if you have a different preference)
default_model = "volary-kimi"
[providers.volary]
name = "volary"
type = "anthropic"
base_url = "https://api.volary.ai/v0/orgs/{ORG_ID}/agents/{AGENT_ID}"
api_key = "{YOUR_TOKEN}"
[models.volary-kimi]
provider = "volary"
model = "kimi"
max_context_size = 128000MCP Server
Also register the Volary MCP server in ~/.kimi/mcp.json:
{
"mcpServers": {
"volary": {
"url": "https://api.volary.ai/v0/orgs/{ORG_ID}/agents/{AGENT_ID}/v0/mcp",
"headers": {
"Authorization": "Bearer {YOUR_TOKEN}"
}
}
}
}GitHub Copilot
VS Code
Run this in your shell to install the Volary VS Code extension and launch VS Code with credentials in the environment – the extension seeds them into secure storage on first activation:
export VOLARY_AGENT_URL="https://api.volary.ai/v0/orgs/{ORG_ID}/agents/{AGENT_ID}" VOLARY_TOKEN="{YOUR_TOKEN}" && code --install-extension Volary.volary && codeCopilot CLI
Already using GitHub Copilot CLI? Register the Volary MCP server:
copilot mcp add --transport http volary --header "Authorization: Bearer {YOUR_TOKEN}" "https://api.volary.ai/v0/orgs/{ORG_ID}/agents/{AGENT_ID}/v0/mcp"Then install the SessionStart and Stop hooks – these upload transcripts and seed your root memory index:
mkdir -p ~/.copilot/hooks && cat > ~/.copilot/hooks/volary.json <<'EOF'
{
"hooks": {
"SessionStart": [
{
"hooks": [
{
"type": "command",
"command": "curl -sfS --max-time 20 -H 'Authorization: Bearer {YOUR_TOKEN}' -H 'Content-Type: application/json' --data-binary @- 'https://api.volary.ai/v0/orgs/{ORG_ID}/agents/{AGENT_ID}/copilot/session-start'"
}
]
}
],
"Stop": [
{
"hooks": [
{
"type": "command",
"command": "TP=$(sed -n 's/.*\"transcript_path\"[[:space:]]*:[[:space:]]*\"\\([^\"]*\\)\".*/\\1/p'); [ -n \"$TP\" ] && [ -f \"$TP\" ] && curl -sfS --max-time 30 -H 'Authorization: Bearer {YOUR_TOKEN}' -H 'Content-Type: application/json' --data-binary @\"$TP\" 'https://api.volary.ai/v0/orgs/{ORG_ID}/agents/{AGENT_ID}/copilot/stop' || true"
}
]
}
]
}
}
EOFGemini
Install the Volary Gemini extension:
gemini extensions install https://github.com/volary-ai/gemini-extensionThen configure the extension with your organisation ID, agent ID, and access token from the dashboard. The setup wizard shows the exact values to copy.
API Provider
For any other tool that speaks the OpenAI or Anthropic API, or that supports MCP, you only need three values from the dashboard:
- Base URL –
https://api.volary.ai/v0/orgs/{ORG_ID}/agents/{AGENT_ID}. Use this as a drop-in replacement forhttps://api.openai.comorhttps://api.anthropic.com. - MCP URL –
https://api.volary.ai/v0/orgs/{ORG_ID}/agents/{AGENT_ID}/v0/mcp. Point any MCP-aware client at this URL. - Access token – pass as
Authorization: Bearer {YOUR_TOKEN}orX-Volary-Api-Key: {YOUR_TOKEN}.
Ready to get started? Sign up and create your first agent, or check the API guide for programmatic integration.