Add requires field to Tool for declaring execution preconditions

Author: Zachary German

docs/specification/draft/changelog.mdx

@@ -18,7 +18,7 @@ N/A
 
 ## Other schema changes
 
-N/A
+1. Add optional `requires` field to `Tool` definitions for declaring opaque execution preconditions.
 
 ## Governance and process updates
 

docs/specification/draft/schema.mdx

@@ -213,6 +213,7 @@ A tool definition includes:
 - `annotations`: Optional properties describing tool behavior
 - `execution`: Optional object describing execution-related properties
   - `taskSupport`: Indicates whether this tool supports [task-augmented execution](/specification/draft/basic/utilities/tasks#tool-level-negotiation). Values: `"forbidden"` (default), `"optional"`, or `"required"`
+  - `requirements`: Optional list of opaque requirement identifiers that MUST be satisfied for the tool to execute successfully (see [Preconditions](#preconditions))
 
 <Warning>
   For trust & safety and security, clients **MUST** consider tool annotations to
@@ -246,6 +247,62 @@ disambiguation.
 
 </Note>
 
+### Preconditions
+
+Tools **MAY** declare an optional `execution.requirements` field — a list of opaque string
+identifiers that represent preconditions for successful execution:
+
+```json
+{
+  "name": "launch_rocket",
+  "description": "Initiates a rocket launch sequence. Only available to authorized mission control personnel when weather conditions are clear and the launch window is open.",
+  "inputSchema": {
+    "type": "object",
+    "properties": {
+      "to": { "type": "string" },
+      "subject": { "type": "string" },
+      "body": { "type": "string" }
+    },
+    "required": ["to", "subject", "body"]
+  },
+  "execution": {
+    "requirements": [
+      "auth:oauth2",
+      "auth:claim:role:mission_control",
+      "capability:rocket.launch",
+      "env:production",
+      "state:weather.clear",
+      "feature:launch_window.open"
+    ]
+  }
+}
+```
+
+Requirement strings are intentionally opaque — MCP does not define or interpret their
+meaning. Strings **SHOULD** be namespaced (e.g., `capability:rocket.launch`) to avoid
+collisions across implementations.
+
+#### Server Behavior
+
+- Servers **MUST** enforce all declared requirements at execution time.
+- If any requirement is not satisfied, the server **MUST** fail the call.
+- Error responses **SHOULD** indicate which requirements were unmet when possible.
+
+#### Client Behavior
+
+Clients **MAY** use `requirements` to influence behavior, subject to the following constraints:
+
+- Clients **MUST** treat requirement strings as opaque.
+- Clients **MUST NOT** infer equivalence between different strings.
+
+- Clients **MUST NOT** use `requirements` for any decision-making unless:
+  - All requirements are known to the client.
+  - The client can determine that at least one requirement cannot be satisfied.
+
+- If the above conditions are met, clients **MAY** use `requirements` to influence behavior (e.g., filtering, ranking, or avoiding failures).
+
+- If any requirement is unknown, or if all requirements may be satisfied, clients **MUST** behave as if `requirements` were not present.
+
 ### Tool Result
 
 Tool results may contain [**structured**](#structured-content) or **unstructured** content.

docs/specification/draft/server/tools.mdx

@@ -0,0 +1,23 @@
+{
+  "name": "launch_rocket",
+  "description": "Initiates a rocket launch sequence. Only available to authorized mission control personnel when weather conditions are clear and the launch window is open.",
+  "inputSchema": {
+    "type": "object",
+    "properties": {
+      "to": { "type": "string" },
+      "subject": { "type": "string" },
+      "body": { "type": "string" }
+    },
+    "required": ["to", "subject", "body"]
+  },
+  "execution": {
+    "requirements": [
+      "auth:oauth2",
+      "auth:claim:role:mission_control",
+      "capability:rocket.launch",
+      "env:production",
+      "state:weather.clear",
+      "feature:launch_window.open"
+    ]
+  }
+}

schema/draft/examples/Tool/with-requirements.json

@@ -1708,6 +1708,28 @@ export interface ToolExecution {
    * Default: `"forbidden"`
    */
   taskSupport?: "forbidden" | "optional" | "required";
+
+  /**
+   * An optional list of requirement identifiers that MUST be satisfied for this tool
+   * to execute successfully.
+   *
+   * Each value is an opaque string. Strings SHOULD be namespaced (e.g.,
+   * `custom:email.send`) to avoid collisions. MCP does not define or interpret
+   * these values.
+   *
+   * Servers MUST enforce all declared requirements at execution time. If any
+   * requirement is not satisfied, the server MUST fail the call.
+   *
+   * Clients MUST treat requirement strings as opaque and MUST NOT infer equivalence
+   * between different strings. Clients MUST NOT use `requirements` for any
+   * decision-making unless all requirements are known to the client and the client
+   * can determine that at least one requirement cannot be satisfied. If those
+   * conditions are met, clients MAY use `requirements` to influence behavior (e.g.,
+   * filtering, ranking, or avoiding tool use). If any requirement is unknown, or if
+   * all requirements may be satisfied, clients MUST behave as if `requirements` were
+   * not present.
+   */
+  requirements?: string[];
 }
 
 /**
@@ -1725,6 +1747,9 @@ export interface ToolExecution {
  * @example With output schema for structured content
  * {@includeCode ./examples/Tool/with-output-schema-for-structured-content.json}
  *
+ * @example With requirements field declaring execution preconditions
+ * {@includeCode ./examples/Tool/with-requirements.json}
+ *
  * @category `tools/list`
  */
 export interface Tool extends BaseMetadata, Icons {