@devries-cameron20 / composio Mirror · read-only
Fork 0

Composio powers 1000+ toolkits, tool search, context management, authentication, and a sandboxed workbench to help you build AI agents that turn intent into action.

Mirrored from https://github.com/Camo67/composio.git

Synced Nov 21, 04:45 UTC

next
mdx 347 lines 15.8 KB
Raw
1 ---
2 title: "Session"
3 description: "A Composio session \u2014 the object returned by `composio.create(...)` / `composio.use(...)`. Use it to fetch session-scoped tools, authorize toolkits,..."
4 ---
5
6 ## Properties
7
8 | Name | Type |
9 |------|------|
10 | `session_id` | `str` |
11 | `config` | `ToolRouterSessionConfig` |
12 | `experimental` | `'ToolRouterSessionExperimental'` |
13 | `config_version` | `int | None` |
14 | `preload` | `Any` |
15
16 ## Methods
17
18 ### tools()
19
20 Get provider-wrapped tools for execution with your AI framework. Returns tools configured for this session, wrapped in the format expected by your AI provider (OpenAI, Anthropic, LangChain, etc.). When custom tools are bound to the session, execution of COMPOSIO_MULTI_EXECUTE_TOOL is intercepted: local tools are executed in-process, remote tools are sent to the backend.
21
22 ```python
23 def tools(modifiers: 'Modifiers' | None = ...) -> TToolCollection
24 ```
25
26 **Parameters**
27
28 | Name | Type |
29 |------|------|
30 | `modifiers?` | `'Modifiers' \| None` |
31
32 **Returns**
33
34 `TToolCollection`
35
36 ---
37
38 ### authorize()
39
40 Authorize a toolkit for the user and get a connection request. Initiates the OAuth flow and returns a ConnectionRequest with redirect URL.
41
42 ```python
43 def authorize(toolkit: str, *, callback_url: str | None = ..., alias: str | None = ..., experimental: session_link_params.Experimental | None = ...) -> ConnectionRequest
44 ```
45
46 **Parameters**
47
48 | Name | Type | Description |
49 |------|------|-------------|
50 | `toolkit` | `str` | |
51 | `callback_url?` | `str \| None` | |
52 | `alias?` | `str \| None` | Human-readable alias for the connection. Must be unique per userId and toolkit within the project. |
53 | `experimental?` | `session_link_params.Experimental \| None` | Experimental options for this connection. Pass an `Experimental` dict with `account_type` and/or `acl_config_for_shared` to create a SHARED connection with a per-user ACL. Experimental — shape may change in future releases. |
54
55 **Returns**
56
57 `ConnectionRequest`
58
59 ---
60
61 ### toolkits()
62
63 Get toolkit connection states for the session.
64
65 ```python
66 def toolkits(*, toolkits: List[str] | None = ..., next_cursor: str | None = ..., limit: int | None = ..., is_connected: bool | None = ..., search: str | None = ...) -> ToolkitConnectionsDetails
67 ```
68
69 **Parameters**
70
71 | Name | Type |
72 |------|------|
73 | `toolkits?` | `List[str] \| None` |
74 | `next_cursor?` | `str \| None` |
75 | `limit?` | `int \| None` |
76 | `is_connected?` | `bool \| None` |
77 | `search?` | `str \| None` |
78
79 **Returns**
80
81 `ToolkitConnectionsDetails`
82
83 ---
84
85 ### search()
86
87 Search for tools by semantic use case. Returns relevant tools for the given query with schemas and guidance.
88
89 ```python
90 def search(*, query: str, model: str | None = ...) -> ToolRouterSessionSearchResponse
91 ```
92
93 **Parameters**
94
95 | Name | Type |
96 |------|------|
97 | `query` | `str` |
98 | `model?` | `str \| None` |
99
100 **Returns**
101
102 `ToolRouterSessionSearchResponse`
103
104 ---
105
106 ### execute()
107
108 Execute a tool within the session. Use `tool_slug=...` (not the direct Tools API's `slug=...`) and `arguments={...}`. The return value is a Pydantic model: read `result.result_type == "completed"` for execution success, `result.model_dump(mode="json")` for a JSON-compatible dictionary, or `result.model_dump_json()` for a JSON string. Unlike provider helper results and `after_execute` modifier dictionaries, it has no `successful` key. This call has no model tool-call ID; retain that ID separately when building provider messages. For custom tools, accepts the full slug (e.g. "LOCAL_GREP") or the original slug (e.g. "GREP") when that original slug is unique across the session's custom tools and toolkits. Custom tools are executed in-process; remote tools are sent to the Composio backend.
109
110 ```python
111 def execute(tool_slug: str, *, arguments: Dict[str, Any] | None = ..., account: str | None = ..., input_responses: Dict[str, ToolRouterInputResponse] | None = ..., request_state: str | None = ...) -> ToolRouterSessionExecuteResponse
112 ```
113
114 **Parameters**
115
116 | Name | Type | Description |
117 |------|------|-------------|
118 | `tool_slug` | `str` | |
119 | `arguments?` | `Dict[str, Any] \| None` | |
120 | `account?` | `str \| None` | Account ID or alias for direct app tool execution in multi-account sessions. Helper/meta tools either ignore this top-level field or define their own account-selection fields. |
121 | `input_responses?` | `Dict[str, ToolRouterInputResponse] \| None` | Experimental. The user's answers to an earlier `ToolInputRequiredError`, keyed by the IDs in its `input_requests`. Repeat the same call (same tool and arguments) with them. An approved call runs; a denied or declined one returns `result_type="failed"`. |
122 | `request_state?` | `str \| None` | Experimental. The `request_state` of that error, sent back unchanged. Both paths return a `ToolRouterSessionExecuteResponse` with `data`, `error`, `log_id`, `result_type`, and `instant_charge` attributes. `result_type` is `"completed"`, `"failed"` or `"blocked"`; a failed execution can carry a `None` or empty `error`, so read `result_type` to tell them apart. A `"blocked"` call (a session hook blocked it) has empty `data`, the hook's message in `error`, and `block_reason` (`type` and `hook_id`). `hook_results` lists every session hook that applied. A log ID correlates an execution; it does not prove that a provider performed a side effect. Inspect the tool-specific `data` as well. Transport and authentication errors raise instead of returning an execution result. Handle authentication recovery separately from returned tool failures; retry/idempotency decisions belong to you. |
123
124 **Returns**
125
126 `ToolRouterSessionExecuteResponse`
127
128 **Raises**
129
130 - `ToolInputRequiredError` — If the tool needs input from the user (for example an approval) before it can run. Nothing was executed.
131 - `ValidationError` — If the API response carries no `result_type`, or one this SDK does not know.
132
133 ---
134
135 ### execute_meta()
136
137 Execute a Composio meta tool (`COMPOSIO_*`) within the session. `execute` also runs meta tools. This method only accepts meta tool slugs, so a typo is a type error instead of a failed request. The call is sent to the Composio backend as is: custom tools named in a `COMPOSIO_MULTI_EXECUTE_TOOL` call are not run in-process here.
138
139 ```python
140 def execute_meta(slug: ToolRouterSessionMetaToolSlug, *, arguments: Dict[str, Any] | None = ..., input_responses: Dict[str, ToolRouterInputResponse] | None = ..., request_state: str | None = ...) -> ToolRouterSessionExecuteResponse
141 ```
142
143 **Parameters**
144
145 | Name | Type | Description |
146 |------|------|-------------|
147 | `slug` | `ToolRouterSessionMetaToolSlug` | The meta tool slug to execute. |
148 | `arguments?` | `Dict[str, Any] \| None` | Optional meta tool arguments. |
149 | `input_responses?` | `Dict[str, ToolRouterInputResponse] \| None` | Experimental. The user's answers to an earlier `ToolInputRequiredError`, keyed by the IDs in its `input_requests`. Repeat the same call with them, as in `execute`. |
150 | `request_state?` | `str \| None` | Experimental. The `request_state` of that error, sent back unchanged. |
151
152 **Returns**
153
154 `ToolRouterSessionExecuteResponse` — The tool execution result, in the same shape as `execute`.
155
156 **Raises**
157
158 - `ToolInputRequiredError` — If the tool needs input from the user (for example an approval) before it can run. Nothing was executed.
159 - `ValidationError` — If the API response carries no `result_type`, or one this SDK does not know.
160
161 **Example**
162
163 ```python
164 result = session.execute_meta(
165 "COMPOSIO_SEARCH_TOOLS",
166 arguments={"queries": [{"use_case": "send an email"}]},
167 )
168 print(result.result_type, result.data)
169 ```
170
171 ---
172
173 ### local_tools()
174
175 List all local tools registered in this session. Returns tools with their final slugs, schemas, and resolved toolkit.
176
177 ```python
178 def local_tools(*, toolkit: str | None = ...) -> List[RegisteredLocalTool]
179 ```
180
181 **Parameters**
182
183 | Name | Type | Description |
184 |------|------|-------------|
185 | `toolkit?` | `str \| None` | Filter by toolkit slug (e.g. 'gmail', 'DEV_TOOLS') |
186
187 **Returns**
188
189 `List[RegisteredLocalTool]` — Array of registered local tools
190
191 ---
192
193 ### local_toolkits()
194
195 List all toolkits of local tools registered in this session. Returns toolkits with their tools showing final slugs.
196
197 ```python
198 def local_toolkits() -> List[RegisteredLocalToolkit]
199 ```
200
201 **Returns**
202
203 `List[RegisteredLocalToolkit]`
204
205 ---
206
207 ### custom_tools()
208
209 Deprecated alias for `local_tools`.
210
211 ```python
212 def custom_tools(*, toolkit: str | None = ...) -> List[RegisteredCustomTool]
213 ```
214
215 **Parameters**
216
217 | Name | Type |
218 |------|------|
219 | `toolkit?` | `str \| None` |
220
221 **Returns**
222
223 `List[RegisteredCustomTool]`
224
225 ---
226
227 ### custom_toolkits()
228
229 Deprecated alias for `local_toolkits`.
230
231 ```python
232 def custom_toolkits() -> List[RegisteredCustomToolkit]
233 ```
234
235 **Returns**
236
237 `List[RegisteredCustomToolkit]`
238
239 ---
240
241 ### proxy_execute()
242
243 Proxy an API call through Composio's auth layer.
244
245 ```python
246 def proxy_execute(*, toolkit: str, endpoint: str, method: Literal['GET', 'POST', 'PUT', 'DELETE', 'PATCH'], body: Any = ..., parameters: List[Dict[str, Any]] | None = ...) -> ToolRouterSessionProxyExecuteResponse
247 ```
248
249 **Parameters**
250
251 | Name | Type | Description |
252 |------|------|-------------|
253 | `toolkit` | `str` | Composio toolkit slug (e.g. 'gmail', 'github') |
254 | `endpoint` | `str` | API endpoint URL |
255 | `method` | `Literal['GET', 'POST', 'PUT', 'DELETE', 'PATCH']` | HTTP method |
256 | `body?` | `Any` | Request body (for POST, PUT, PATCH) |
257 | `parameters?` | `List[Dict[str, Any]] \| None` | Query/header parameters |
258
259 **Returns**
260
261 `ToolRouterSessionProxyExecuteResponse` — The proxied API response with `result_type="completed"`, or `result_type="blocked"` with the hook's `message` when a session hook blocked the request (it then has no `status`)
262
263 **Raises**
264
265 - `ToolInputRequiredError` — If the call needs input from the user before it can run. Nothing was executed.
266
267 ---
268
269 ### update()
270
271 Partially update the session configuration. Only the fields provided are changed; omitted fields are preserved. For each policy block `None` removes the stored override (which can increase access: `toolkits=None` restores the unrestricted default, while `toolkits={"enable": []}` denies every app toolkit and is sent as-is). Supplied `tools`, `auth_configs` and `connected_accounts` maps replace the stored map entirely. Inside `manage_connections`, `callback_url=None` removes only the stored callback URL. Experimental `instant` accepts `False` to disable billed access or an object to set its filters; it does not accept `None`. Any object, even one that only sets `return_instant_charge` or `return_prices_in_search`, re-enables Instant usage on a Session set to `False`. `proxy_execute` takes a bool or `{"enable": bool}`; `True` exposes the `COMPOSIO_PROXY_EXECUTE` meta tool, `False` also turns off the sandbox `proxy_execute()` helper, and `None` removes the stored block. `hooks` replaces the whole list of hooks, and `None` removes every hook. `hooks` cannot be combined with `experimental={'session_config_id': ...}` (`hooks=None` counts). By default the request carries no precondition: the last writer wins. Pass `expected_config_version` (for example this object's `config_version`) to make the update conditional: the API then applies it only when the stored version still matches, and a concurrent change raises `SessionConfigConflictError` (HTTP 409) instead of being overwritten. The API must support the `expected_config_version` field; otherwise it rejects the request with a 400. `expected_config_version=False` is the same as omitting it. The PATCH is never retried by the transport, so a 409 is reported exactly once. On conflict this object stays unchanged: re-fetch the session with `composio.sessions.use(session_id)` and retry against the fresh `config_version`. `config`, `config_version` and `preload` are refreshed in place only after a successful response, and the updated `config` is returned. `workbench` is a backwards-compatible alias for `sandbox`. Prefer `sandbox` in new code. All other parameters use the same types as the generated `client.tool_router.session.patch()` method.
272
273 ```python
274 def update(*, toolkits: Union[session_patch_params.Toolkits | None, 'Omit'] = ..., instant: Union[Literal[False], ToolRouterInstantConfig, 'Omit'] = ..., tools: Union[Dict[str, session_patch_params.Tools] | None, 'Omit'] = ..., tags: Union[session_patch_params.Tags | None, 'Omit'] = ..., auth_configs: Union[Dict[str, str] | None, 'Omit'] = ..., connected_accounts: Union[Dict[str, SequenceNotStr[str]] | None, 'Omit'] = ..., manage_connections: Union[session_patch_params.ManageConnections | None, ToolRouterUpdateManageConnectionsConfig | None, 'Omit'] = ..., sandbox: Union[session_patch_params.Workbench | None, 'Omit'] = ..., workbench: Union[session_patch_params.Workbench | None, 'Omit'] = ..., multi_account: Union[session_patch_params.MultiAccount | None, ToolRouterUpdateMultiAccountConfig | None, 'Omit'] = ..., proxy_execute: Union[Union[bool, ToolRouterProxyExecuteConfig] | None, 'Omit'] = ..., hooks: Union[List[ToolRouterHook] | None, 'Omit'] = ..., preload: Union[session_patch_params.Preload | None, 'Omit'] = ..., search: Union[session_patch_params.Search | None, 'Omit'] = ..., execute: Union[session_patch_params.Execute | None, 'Omit'] = ..., experimental: Union[session_patch_params.Experimental | None, ToolRouterUpdateExperimentalConfig | None, 'Omit'] = ..., expected_config_version: Union[int, None, Literal[False]] = ...) -> ToolRouterSessionConfig
275 ```
276
277 **Parameters**
278
279 | Name | Type |
280 |------|------|
281 | `toolkits?` | `Union[session_patch_params.Toolkits \| None, 'Omit']` |
282 | `instant?` | `Union[Literal[False], ToolRouterInstantConfig, 'Omit']` |
283 | `tools?` | `Union[Dict[str, session_patch_params.Tools] \| None, 'Omit']` |
284 | `tags?` | `Union[session_patch_params.Tags \| None, 'Omit']` |
285 | `auth_configs?` | `Union[Dict[str, str] \| None, 'Omit']` |
286 | `connected_accounts?` | `Union[Dict[str, SequenceNotStr[str]] \| None, 'Omit']` |
287 | `manage_connections?` | `Union[session_patch_params.ManageConnections \| None, ToolRouterUpdateManageConnectionsConfig \| None, 'Omit']` |
288 | `sandbox?` | `Union[session_patch_params.Workbench \| None, 'Omit']` |
289 | `workbench?` | `Union[session_patch_params.Workbench \| None, 'Omit']` |
290 | `multi_account?` | `Union[session_patch_params.MultiAccount \| None, ToolRouterUpdateMultiAccountConfig \| None, 'Omit']` |
291 | `proxy_execute?` | `Union[Union[bool, ToolRouterProxyExecuteConfig] \| None, 'Omit']` |
292 | `hooks?` | `Union[List[ToolRouterHook] \| None, 'Omit']` |
293 | `preload?` | `Union[session_patch_params.Preload \| None, 'Omit']` |
294 | `search?` | `Union[session_patch_params.Search \| None, 'Omit']` |
295 | `execute?` | `Union[session_patch_params.Execute \| None, 'Omit']` |
296 | `experimental?` | `Union[session_patch_params.Experimental \| None, ToolRouterUpdateExperimentalConfig \| None, 'Omit']` |
297 | `expected_config_version?` | `Union[int, None, Literal[False]]` |
298
299 **Returns**
300
301 `ToolRouterSessionConfig`
302
303 ---
304
305 ### list_config_history()
306
307 List the configuration history of this session, newest first. Every `update()` records a new config version; this returns those versions with cursor-based pagination.
308
309 ```python
310 def list_config_history(**query: Unpack[session_config_history_params.SessionConfigHistoryParams]) -> session_config_history_response.SessionConfigHistoryResponse
311 ```
312
313 **Parameters**
314
315 | Name | Type |
316 |------|------|
317 | `query?` | `Unpack[session_config_history_params.SessionConfigHistoryParams]` |
318
319 **Returns**
320
321 `session_config_history_response.SessionConfigHistoryResponse` — The config versions under `.items` plus pagination fields.
322
323 **Example**
324
325 ```python
326 history = session.list_config_history(limit=10)
327 for entry in history.items:
328 print(entry.version, entry.is_current)
329 ```
330
331 ---
332
333 ### delete()
334
335 Delete this session. Deleted sessions immediately stop being retrievable or executable. An already-deleted session surfaces the backend 404.
336
337 ```python
338 def delete() -> ToolRouterSessionDeleteResponse
339 ```
340
341 **Returns**
342
343 `ToolRouterSessionDeleteResponse`
344
345 ---
346
347 [View source](https://github.com/composiohq/composio/blob/next/python/composio/core/models/tool_router_session.py#L316)