Skip to content
Documentation
Connect a clientCline

Cline

Set up Cline: choose the API format, Guard URL, key and model.

Set up with your own AI
Model protocol

Claude, Kimi, Composer and Grok

Cline is a popular agent extension for VS Code. It connects to Guard via the Anthropic provider and accepts a third-party key with no paid restrictions (unlike Cursor).

  1. Install the Cline extension from the VS Code Marketplace.
  2. In Cline settings → API Provider → choose Anthropic.
  3. Enable Use custom base URL and enter https://api.guardrelay.ai.
  4. API Key — your gd-… from the bot.
  5. Model — claude-sonnet-5-5 or claude-opus-5.
FieldValue
API ProviderAnthropic
Custom Base URLhttps://api.guardrelay.ai
API Keygd-…
Modelclaude-sonnet-5-5 / claude-opus-5

Kimi

For Kimi in Cline, select k3 or k3-256k. With Anthropic use Custom Base URL https://api.guardrelay.ai and your gd-… key; with an OpenAI-compatible provider use the fields below. In Cline, kimi-for-coding and kimi-for-coding-highspeed may return 403 kimi_agent_not_allowed: use a supported client such as Roo Code or Kilo Code. Do not remove or impersonate the agent name in the system prompt to bypass this restriction.

FieldValue
API ProviderOpenAI-compatible
Base URLhttps://api.guardrelay.ai/v1
API Keygd-…
Modelk3, k3-256k

Composer and Grok

They work through the Anthropic provider (https://api.guardrelay.ai) and through an OpenAI-compatible one (https://api.guardrelay.ai/v1). Put the exact id in Model, for example cursor-grok-4.6-high or composer-2.5.

GLM 5.3 in Cline

Install and enable the legacy engine

  1. Install VS Code and Cline extension version 4.1.20 from Marketplace.
  2. Open Preferences: Open User Settings (JSON), add the field below, and restart the VS Code window.
  3. Open a test folder through File → Open Folder.
json
{
  "cline.rollout.bundleOverride": "legacy"
}

Get a Guard key

Get a universal gd-… key from the direct API link in @guardapi and choose an API package. Do not put the key in a project, repository, task prompt, or screenshot. Balance and usage are available in your Guard account.

Configure OpenAI Compatible

  1. Select the gear in Cline and choose OpenAI Compatible under API Provider.
  2. Set Base URL to https://api.guardrelay.ai/glm/v1; Cline appends /chat/completions itself.
  3. Enter the gd-… key in API Key and manually set Model ID to glm-5.3.
  4. Leave Reasoning Effort and other undocumented fields disabled. If the UI requires an output cap, use a conservative client budget of 4096, not an assumed model limit.
Cline fieldValue
API ProviderOpenAI Compatible
Base URLhttps://api.guardrelay.ai/glm/v1
API Keygd-…
Model IDglm-5.3
Final requestPOST https://api.guardrelay.ai/glm/v1/chat/completions

Quick check and continuation

  1. Create a test README.md containing GUARD_GLM_OK and open a new Cline task.
  2. Send: Read README.md and reply with exactly GUARD_GLM_OK. Change nothing; approve only file reading.
  3. In the same task, ask for the line again without rereading the file. The answer should retain context.

Models and confirmed capabilities

Model IDWhen to use it
glm-5.3Default model; text and coding tasks.
glm-5.3-flashFast model; at API level it accepts PNG and PDF through all three protocols, and MP4 through Chat Completions only.
  • Streaming Chat, client-side file and terminal tools, and continuation of one task are confirmed.
  • The client itself executes file, edit, and terminal tools. This profile does not connect Z.AI tools or MCP automatically.
  • The Chat API permits MP4 with Flash and PNG/PDF at API level; Cline attachment buttons were not checked.
  • The profile uses OpenAI-compatible Chat Completions. The GLM server prompt is off by default for ordinary API access; client setup does not enable it.
  • Native TXT, DOCX, and XLSX are outside the confirmed GLM contract. Use PDF for documents, or PNG with glm-5.3-flash for images.

If the connection fails

SymptomWhat to check
401 invalid_api_keyThe client received the complete active gd-… key; the variable is set in the same process that starts the client.
404 model_not_foundThe model ID is exactly glm-5.3 or glm-5.3-flash; old GLM IDs, aliases, Turbo, and FlashX are unsupported.
404 or incorrect pathBase URL is https://api.guardrelay.ai/glm/v1. The client appends /chat/completions; do not add it twice.
400 invalid_requestThe documented API format is selected; reasoning, temperature, MCP, or other unconfirmed fields were not copied from an older profile.
402 insufficient_balanceCheck available balance and unfinished reservations in your Guard account.
The request uses another engineCheck Cline 4.1.20 and the user setting cline.rollout.bundleOverride = legacy; SDK-next compatibility is not claimed by this profile.

Gemini in Cline

Install and enable the legacy engine

  1. Install VS Code and the Cline extension from Marketplace.
  2. Open Preferences: Open User Settings (JSON), add the field below, and restart the VS Code window.
  3. Open a test folder through File → Open Folder.
json
{
  "cline.rollout.bundleOverride": "legacy"
}

Get a Guard key

Get a universal gd-… key from the direct API link in @guardapi and choose an API package. Do not put the key in a project, repository, task prompt, or screenshot. Balance and usage are available in your Guard account.

Configure OpenAI Compatible

  1. Select the gear in Cline and choose OpenAI Compatible under API Provider.
  2. Set Base URL to https://api.guardrelay.ai/v1; Cline appends /chat/completions itself.
  3. Enter the gd-… key in API Key and manually set Model ID to gemini-4-argon or another ID from the table below.
  4. Leave Reasoning Effort and other undocumented fields disabled. If the UI requires an output cap, use a conservative budget of 4096.
Cline fieldValue
API ProviderOpenAI Compatible
Base URLhttps://api.guardrelay.ai/v1
API Keygd-…
Model IDgemini-4-argon
Final requestPOST https://api.guardrelay.ai/v1/chat/completions

Quick check and continuation

  1. Create a test README.md containing GUARD_GEMINI_OK and open a new Cline task.
  2. Send: Read README.md and reply with exactly GUARD_GEMINI_OK. Change nothing; approve only file reading.
  3. In the same task, ask for the line again without rereading the file. The answer should retain context.

Gemini models and capabilities

Model IDInput and window
gemini-4-argonText, images, PDF; 1,048,576-token window, shared by input and answer.
gemini-3.8-flashText, images, PDF; 1,048,576-token window for input.
  • Cline itself executes file and terminal tools; it sends the task history to Gemini with every request.
  • The Guard catalog takes text, images, and PDF as input and returns text. How to send a file or an image depends on the client's interface; this guide does not describe attachments.
  • The profile uses OpenAI-compatible Chat Completions. The server prompt is a separate setting of the Gemini family: it applies only when the key has the entitlement and the option is on; client setup does not enable it.

If the connection fails

SymptomWhat to check
401 invalid_api_keyThe client received the complete active gd-… key; the variable is set in the same process that starts the client.
404 model_not_foundThe model ID is an exact ID from the table above or the short name gemini; other spellings are unsupported.
404 or incorrect pathBase URL is https://api.guardrelay.ai/v1. The client appends /chat/completions; do not add it twice.
400 invalid_requestThe documented API format is selected; extra settings from an older profile were not copied: reasoning, temperature, MCP, and others.
402 insufficient_creditCheck available balance and unfinished reservations in your Guard account.
The request uses another engineCheck the user setting cline.rollout.bundleOverride = legacy; other engines are not described by this profile.

Muse in Cline

Install and enable the legacy engine

  1. Install VS Code and the Cline extension from Marketplace.
  2. Open Preferences: Open User Settings (JSON), add the field below, and restart the VS Code window.
  3. Open a test folder through File → Open Folder.
json
{
  "cline.rollout.bundleOverride": "legacy"
}

Get a Guard key

Get a universal gd-… key from the direct API link in @guardapi and choose an API package. Do not put the key in a project, repository, task prompt, or screenshot. Balance and usage are available in your Guard account.

Configure OpenAI Compatible

  1. Select the gear in Cline and choose OpenAI Compatible under API Provider.
  2. Set Base URL to https://api.guardrelay.ai/v1; Cline appends /chat/completions itself.
  3. Enter the gd-… key in API Key and manually set Model ID to muse-wonder-1.0 or another ID from the table below.
  4. Leave Reasoning Effort and other undocumented fields disabled. If the UI requires an output cap, use a conservative budget of 4096.
Cline fieldValue
API ProviderOpenAI Compatible
Base URLhttps://api.guardrelay.ai/v1
API Keygd-…
Model IDmuse-wonder-1.0
Final requestPOST https://api.guardrelay.ai/v1/chat/completions

Quick check and continuation

  1. Create a test README.md containing GUARD_MUSE_OK and open a new Cline task.
  2. Send: Read README.md and reply with exactly GUARD_MUSE_OK. Change nothing; approve only file reading.
  3. In the same task, ask for the line again without rereading the file. The answer should retain context.

Muse models and capabilities

Model IDInput and window
muse-wonder-1.0Text, images, PDF; 1,048,576-token window.
muse-spark-1.3Text, images, PDF; 1,048,576-token window.
muse-spark-1.2Text, images, PDF; 1,048,576-token window.
  • Cline itself executes file and terminal tools; it sends the task history to Muse with every request.
  • The Guard catalog takes text, images, and PDF as input and returns text. How to send a file or an image depends on the client's interface; this guide does not describe attachments.
  • The profile uses OpenAI-compatible Chat Completions. The server prompt is a separate setting of the Muse family: it applies only when the key has the entitlement and the option is on; client setup does not enable it.

If the connection fails

SymptomWhat to check
401 invalid_api_keyThe client received the complete active gd-… key; the variable is set in the same process that starts the client.
404 model_not_foundThe model ID is an exact ID from the table above or the short name muse; other spellings are unsupported.
404 or incorrect pathBase URL is https://api.guardrelay.ai/v1. The client appends /chat/completions; do not add it twice.
400 invalid_requestThe documented API format is selected; extra settings from an older profile were not copied: reasoning, temperature, MCP, and others.
402 insufficient_creditCheck available balance and unfinished reservations in your Guard account.
The request uses another engineCheck the user setting cline.rollout.bundleOverride = legacy; other engines are not described by this profile.

Codex in Cline

For Cline 4.1.20, use the legacy engine and OpenAI Compatible: this profile sends streaming Chat Completions to the Codex family address and carries history within the same task. Get your gd-… key from the Guard bot. This profile makes no SDK-next compatibility claim.

For a connection check, use a plain chat without files and ask “Reply: connection established”. To check project work, create a separate guard-test folder in File Explorer or Finder, open it in your editor, and create README.md containing “This is a Guard test project”. The file-reading example below uses that folder.

Model IDName
gpt-6-astraGPT-6 Astra
gpt-6.1-solGPT-6.1 Sol
gpt-6-solGPT-6 Sol
gpt-6-lunaGPT-6 Luna
gpt-5.6-solGPT-5.6 Sol
gpt-5.6-terraGPT-5.6 Terra
gpt-5.6-lunaGPT-5.6 Luna

Install and connect

  1. Install Cline 4.1.20. In VS Code User Settings JSON add "cline.rollout.bundleOverride": "legacy", restart the window, and open a test folder without secrets and with a README.md through File → Open Folder.
  2. Select the settings gear in Cline and choose OpenAI Compatible under API Provider. A Cline account sign-in is not needed for your own Guard key.
  3. Set Base URL to https://api.guardrelay.ai/v1.
  4. Paste your gd-… key from the bot into API Key. Do not put it in a project file or task message.
  5. Choose custom model ID entry under Model ID and enter exactly gpt-6-astra. If the model list is empty, type the ID manually. Other exact IDs are under Models.
  6. Leave optional Model Configuration numeric limits unchanged. If Reasoning Effort is available, choose only a value listed for that model under Models; otherwise keep its default. Return to the Cline panel.
  7. Start a new task and send: Read README.md and explain in two sentences what this test project does. Do not change anything. If Cline asks for permission, allow only a read of that file.
  8. After the reply, send in the same task: Name one fact from README.md that supports your explanation. Start a new task for a different question.
Cline fieldValue
API ProviderOpenAI Compatible
Base URLhttps://api.guardrelay.ai/v1
API Keygd-…
Model IDgpt-6-astra
Final requestPOST https://api.guardrelay.ai/v1/chat/completions

After checking file reading, you can request a change: “Add a Test section and one descriptive line to README.md.” Review the proposed edit and approve it if it matches your task. Command, file and tool permissions are controlled by the client; connecting Guard does not require granting unrestricted access to your whole computer.

Sections

On this page