Skip to content

Add openrouter:advisor server tool to skills #47

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
devin-ai-integration wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Add openrouter:advisor server tool to skills #47

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions skills/create-agent-tui/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -118,6 +118,7 @@ The skill presents an interactive checklist when invoked. You pick what you need
| Web Search | on | Real-time web search via `openrouter:web_search` |
| Datetime | on | Current date/time via `openrouter:datetime` |
| Image Generation | off | Generate images via `openrouter:image_generation` |
| Advisor | off | Consult a stronger model mid-generation via `openrouter:advisor` |

### User-defined tools (your code, executed locally)

Expand Down
6 changes: 6 additions & 0 deletions skills/create-agent-tui/SKILL.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -42,9 +42,12 @@ Present this as a multi-select checklist. Items marked **ON** are pre-selected d
| Web Search | `openrouter:web_search` | ON | engine, max_results, domain filtering |
| Datetime | `openrouter:datetime` | ON | timezone |
| Image Generation | `openrouter:image_generation` | OFF | model, quality, size, format |
| Advisor | `openrouter:advisor` | OFF | model, instructions, tools, forward_transcript, named advisors |

Server tools go in the `tools` array alongside user-defined tools. No client code needed — OpenRouter executes them.

**Advisor** lets the agent consult a stronger model mid-generation — before committing to an approach, when stuck, or before declaring done. Pass `parameters.model` to pin an advisor model, or let the executor choose per call. Supports `instructions`, `forward_transcript`, sub-agent `tools`, and named advisor profiles via `parameters.advisors`. Docs: [openrouter.ai/docs/guides/features/server-tools/advisor](https://openrouter.ai/docs/guides/features/server-tools/advisor).

### User-Defined Tools (client-side, generated into src/tools/)

| Tool | Default | Description |
Expand Down Expand Up @@ -347,6 +350,9 @@ export const tools = [
// Server tools — executed by OpenRouter, no client implementation needed
serverTool({ type: 'openrouter:web_search' }),
serverTool({ type: 'openrouter:datetime', parameters: { timezone: 'UTC' } }),

// Advisor — consult a stronger model mid-generation (off by default, add when needed)
// serverTool({ type: 'openrouter:advisor', parameters: { model: '~anthropic/claude-opus-latest' } }),
];
```

Expand Down
3 changes: 3 additions & 0 deletions skills/create-agent-tui/sample/src/tools/index.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -18,4 +18,7 @@ export const tools = [

serverTool({ type: 'openrouter:web_search' }),
serverTool({ type: 'openrouter:datetime', parameters: { timezone: 'UTC' } }),

// Advisor — consult a stronger model mid-generation (uncomment to enable)
// serverTool({ type: 'openrouter:advisor', parameters: { model: '~anthropic/claude-opus-latest' } }),
];
3 changes: 2 additions & 1 deletion skills/create-headless-agent/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -58,6 +58,7 @@ The skill presents an interactive checklist when invoked:
| Web Fetch | on | Fetch text content from a URL via `openrouter:web_fetch` |
| Datetime | on | Current date/time via `openrouter:datetime` |
| Image Generation | off | Generate images via `openrouter:image_generation` |
| Advisor | off | Consult a stronger model mid-generation via `openrouter:advisor` |

### User-defined tools (your code, executed locally)

Expand Down Expand Up @@ -127,7 +128,7 @@ my-agent/
agent.test.ts Example test (bun:test)
```

Server tools (`openrouter:web_search`, `openrouter:web_fetch`, `openrouter:datetime`) are wired in `tools/index.ts` and execute on OpenRouter's side — no client code.
Server tools (`openrouter:web_search`, `openrouter:web_fetch`, `openrouter:datetime`, `openrouter:advisor`) are wired in `tools/index.ts` and execute on OpenRouter's side — no client code.

## Sample

Expand Down
6 changes: 6 additions & 0 deletions skills/create-headless-agent/SKILL.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -47,9 +47,12 @@ Present this as a multi-select checklist. Items marked **ON** are pre-selected d
| Web Fetch | `openrouter:web_fetch` | ON |
| Datetime | `openrouter:datetime` | ON |
| Image Generation | `openrouter:image_generation` | OFF |
| Advisor | `openrouter:advisor` | OFF |

Server tools go in the `tools` array alongside user-defined tools. No client code needed — OpenRouter executes them. Docs: [openrouter.ai/docs/guides/features/server-tools](https://openrouter.ai/docs/guides/features/server-tools/overview).

**Advisor** lets the agent consult a stronger model mid-generation — before committing to an approach, when stuck, or before declaring done. Pass `parameters.model` to pin an advisor model, or let the executor choose per call. Supports `instructions`, `forward_transcript`, sub-agent `tools`, and named advisor profiles via `parameters.advisors`. Docs: [openrouter.ai/docs/guides/features/server-tools/advisor](https://openrouter.ai/docs/guides/features/server-tools/advisor).

### User-Defined Tools (client-side, generated into src/tools/)

| Tool | Default | Description |
Expand Down Expand Up @@ -355,6 +358,9 @@ export const tools = [
serverTool({ type: 'openrouter:web_search' }),
serverTool({ type: 'openrouter:web_fetch' }),
serverTool({ type: 'openrouter:datetime', parameters: { timezone: 'UTC' } }),

// Advisor — consult a stronger model mid-generation (off by default, add when needed)
// serverTool({ type: 'openrouter:advisor', parameters: { model: '~anthropic/claude-opus-latest' } }),
] as const;
```

Expand Down
3 changes: 3 additions & 0 deletions skills/create-headless-agent/sample/src/tools/index.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -23,4 +23,7 @@ export const tools = [
serverTool({ type: 'openrouter:web_search' }),
serverTool({ type: 'openrouter:web_fetch' }),
serverTool({ type: 'openrouter:datetime', parameters: { timezone: 'UTC' } }),

// Advisor — consult a stronger model mid-generation (uncomment to enable)
// serverTool({ type: 'openrouter:advisor', parameters: { model: '~anthropic/claude-opus-latest' } }),
] as const;
87 changes: 87 additions & 0 deletions skills/openrouter-typescript-sdk/SKILL.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -498,6 +498,93 @@ const manualTool = tool({
});
```

#### Server Tools
OpenRouter provides built-in server tools that execute on OpenRouter's side — no client implementation needed. Use `serverTool()` to add them to the `tools` array alongside regular tools:

```typescript
import { serverTool } from '@openrouter/agent';

const result = client.callModel({
model: 'openai/gpt-5-nano',
input: 'Research the latest on quantum computing',
tools: [
serverTool({ type: 'openrouter:web_search' }),
serverTool({ type: 'openrouter:web_fetch' }),
serverTool({ type: 'openrouter:datetime', parameters: { timezone: 'UTC' } }),
serverTool({ type: 'openrouter:image_generation' }),
serverTool({ type: 'openrouter:advisor', parameters: { model: '~anthropic/claude-opus-latest' } }),
]
});
```

Available server tools: `openrouter:web_search`, `openrouter:web_fetch`, `openrouter:datetime`, `openrouter:image_generation`, `openrouter:advisor`. Docs: [openrouter.ai/docs/guides/features/server-tools](https://openrouter.ai/docs/guides/features/server-tools/overview).

#### Advisor Tool

The `openrouter:advisor` server tool lets a model consult a stronger advisor model mid-generation. The model invokes it with a `prompt` when it hits a decision point, is stuck, or wants validation before proceeding.

```typescript
// Basic — pin an advisor model
serverTool({
type: 'openrouter:advisor',
parameters: { model: '~anthropic/claude-opus-latest' }
})

// With instructions and sub-agent tools
serverTool({
type: 'openrouter:advisor',
parameters: {
model: '~anthropic/claude-opus-latest',
instructions: 'You are a senior staff engineer. Be decisive.',
tools: [{ type: 'openrouter:web_search' }],
forward_transcript: false,
}
})

// Named advisor profiles — the model selects by name per call
serverTool({
type: 'openrouter:advisor',
parameters: {
advisors: [
{
name: 'reviewer',
model: '~anthropic/claude-opus-latest',
instructions: 'You are a critical code reviewer. Find the flaws.',
},
{
name: 'architect',
model: '~openai/gpt-latest',

@perry-the-pr-maintainer perry-the-pr-maintainer Bot Jun 9, 2026

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

[suggestion] ~openai/gpt-latest — unverified OpenRouter alias; this slug does not appear elsewhere in the codebase

Details

The ~ prefix is correct for family aliases per the repo's conventions, but gpt-latest is not used anywhere else in this repo. Existing examples use pinned slugs like openai/gpt-5.2 or openai/gpt-4.1-mini. If ~openai/gpt-latest is a real OpenRouter alias, this is fine — but consider verifying against the models catalogue or using a known alias like ~openai/gpt-4o-latest so developers copy-pasting this example don't hit a routing error.

Prompt for agents 
In `skills/openrouter-typescript-sdk/SKILL.md` at line 556, the named-advisor example uses `model: '~openai/gpt-latest'`. Verify whether this is a valid OpenRouter model alias at https://openrouter.ai/models. If it is not valid, replace it with a confirmed alias (e.g. `~openai/gpt-4o-latest`) or a pinned slug that matches the repo's existing examples (e.g. `openai/gpt-4.1-mini`).

Reviewed at fe4e7d6

instructions: 'You are a systems architect. Think about scale.',
},
]
}
})
```

**Parameters** (`parameters` object on the tool entry):

| Field | Default | Description |
|-------|---------|-------------|
| `model` | Outer request model | Advisor model to consult (any OpenRouter model) |
| `advisors` | None | Roster of named advisor profiles — model selects by `name` per call |
| `instructions` | None | System instructions for the advisor |
| `tools` | None | Tools available to the advisor sub-agent (e.g. `openrouter:web_search`) |
| `forward_transcript` | `false` | Forward full parent conversation to the advisor |
| `max_tool_calls` | Provider default | Max tool-calling steps for the advisor (1–25) |
| `max_completion_tokens` | Provider default | Max output tokens for the advisor call |
| `reasoning` | Provider default | Reasoning config (`{ effort, max_tokens }`) |
| `temperature` | Provider default | Sampling temperature (0–2) |

**Tool-call arguments** (passed by the model when invoking):

| Argument | Description |
|----------|-------------|
| `prompt` | What the model wants advice on (required unless `forward_transcript` is `true`) |
| `name` | Which named advisor profile to consult (only when `advisors` roster is configured) |
| `model` | Advisor model override (only honored when neither the profile nor definition fixes one) |

Docs: [openrouter.ai/docs/guides/features/server-tools/advisor](https://openrouter.ai/docs/guides/features/server-tools/advisor).

---

## Multi-Turn Conversations with Stop Conditions
Expand Down