Hweinstock · Hweinstock · Sep 23, 2026 · Sep 23, 2026 · Sep 23, 2026 · Sep 23, 2026
diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
@@ -85,7 +85,7 @@ npm i -g "./$TARBALL"
 - **The CLI looks frozen in a PowerShell window**: legacy conhost pauses all
   output while text is selected (the title bar shows `Select`). Press `Esc`.
   Windows Terminal does not do this.
-- **`project create` refuses a long path**: Windows caps paths at 260 characters
+- **`agentcore create` refuses a long path**: Windows caps paths at 260 characters
   unless `LongPathsEnabled` is set, and the CDK app's `node_modules` puts its
   deepest file 155 characters below the project root (aws-cdk-lib's own shipped
   fixtures), so the project root must be at most 104 characters. Create the

diff --git a/README.md b/README.md
@@ -29,51 +29,51 @@ those details so you can create, deploy, and invoke agents from your terminal.
 Create a managed Harness project, deploy it, and send a prompt:
 
 ```bash
-agentcore project create --name MyAssistant
+agentcore create --name MyAssistant
 cd MyAssistant
-agentcore project deploy
-agentcore project invoke harness --prompt "Hey, what can you do for me?"
+agentcore deploy
+agentcore invoke harness --prompt "Hey, what can you do for me?"
 ```
 
 To start with code you own instead, create a Runtime project from a template.
 Run this alternative from outside an existing project:
 
 ```bash
-agentcore project create --name MyAgent --template agent-python-strands
+agentcore create --name MyAgent --template agent-python-strands
 ```
 
 ## Command Surface
 
-`project` commands manage local project specifications and their deployments.
-Resource commands operate on deployed resources without requiring a local project.
-
-| Command    | Purpose                                                                       |
-| ---------- | ----------------------------------------------------------------------------- |
-| `project`  | Create, develop, build, deploy, invoke, and inspect a project                 |
-| `harness`  | Manage Harnesses, versions, and endpoints; invoke and inspect them            |
-| `identity` | Manage credential providers                                                   |
-| `runtime`  | Inspect, invoke, and open a shell in deployed Runtimes                        |
-| `memory`   | Inspect Memories, actors, sessions, events, and records                       |
-| `gateway`  | Inspect and invoke Gateways, inspect targets and rules, and generate policies |
-| `payment`  | Inspect payment managers, connectors, sessions, instruments, and balances     |
-| `eval`     | Evaluate agents, manage datasets and configurations, and run experiments      |
-| `feedback` | Submit feedback                                                               |
-| `config`   | Read and write global CLI settings                                            |
-| `update`   | Check for and install CLI updates                                             |
+Project commands manage local project specifications and their deployments. Resource commands
+operate on deployed resources without requiring a local project.
+
+| Command                                                                                            | Purpose                                                                       |
+| -------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------- |
+| `create`, `add`, `export`, `remove`, `dev`, `deploy`, `invoke`, `log`, `traces`, `status`, `build` | Create, develop, build, deploy, invoke, and inspect a project                 |
+| `harness`                                                                                          | Manage Harnesses, versions, and endpoints; invoke and inspect them            |
+| `identity`                                                                                         | Manage credential providers                                                   |
+| `runtime`                                                                                          | Inspect, invoke, and open a shell in deployed Runtimes                        |
+| `memory`                                                                                           | Inspect Memories, actors, sessions, events, and records                       |
+| `gateway`                                                                                          | Inspect and invoke Gateways, inspect targets and rules, and generate policies |
+| `payment`                                                                                          | Inspect payment managers, connectors, sessions, instruments, and balances     |
+| `eval`                                                                                             | Evaluate agents, manage datasets and configurations, and run experiments      |
+| `feedback`                                                                                         | Submit feedback                                                               |
+| `config`                                                                                           | Read and write global CLI settings                                            |
+| `update`                                                                                           | Check for and install CLI updates                                             |
 
 Use `--help` for subcommands and flags, or browse the [command reference](command.md):
 
 ```bash
 agentcore --help
-agentcore project --help
+agentcore add --help
 agentcore runtime invoke --help
 ```
 
 Supported bare commands open their interactive flows in a terminal. Operation
 flags select headless behavior for most commands, but invoke commands can use
 selectors such as `--id` and `--session-id` to seed an interactive console.
-Run `agentcore project create` for guided setup. To create a default project
-without the wizard, run `agentcore project create --name MyAssistant`.
+Run `agentcore create` for guided setup. To create a default project without
+the wizard, run `agentcore create --name MyAssistant`.
 
 Global flags (declared at the root, available on every command):
 
@@ -129,14 +129,14 @@ export class AgentCoreStack extends Stack {
 
 For a Harness, use `this.application.harness("<name>")` instead.
 
-Run `agentcore project deploy` to apply your changes. The names passed to
+Run `agentcore deploy` to apply your changes. The names passed to
 `runtime()` or `harness()` must match the names in your project.
 
 If you use an existing execution role through `executionRoleArn`, CDK cannot
 change its permissions. You'll need to add the required permissions to that role
 yourself.
 
-Note that `agentcore project status` reports only the resources `agentcore.json`
+Note that `agentcore status` reports only the resources `agentcore.json`
 declares, not the ones you add in the stack.
 
 ## Documentation

diff --git a/docs/harness-project-configuration.md b/docs/harness-project-configuration.md
@@ -17,7 +17,7 @@
 
 ## Harness Project Files
 
-`project create` (without `--template`) and `project add harness` share the same
+`agentcore create` (without `--template`) and `agentcore add harness` share the same
 scaffolding flow. Each harness has `app/<name>/harness.yaml` and
 `app/<name>/system-prompt.md`:
 
@@ -50,7 +50,7 @@ memory:
   mode: managed
 ```
 
-Edit the file, then run `agentcore project deploy` from the project directory to
+Edit the file, then run `agentcore deploy` from the project directory to
 apply changes. A local edit does not update an already deployed harness.
 The examples below are separate alternatives or sections to add to your file.
 Replace a section when switching modes rather than keeping fields from both.
@@ -151,7 +151,7 @@ systemPrompt: |
 Path-shaped inline values ending in `.md` or `.txt` are rejected by the current
 schema. Use the conventional `system-prompt.md` file instead.
 
-`project add harness --system-prompt "Your instructions"` writes the supplied
+`agentcore add harness --system-prompt "Your instructions"` writes the supplied
 text to `system-prompt.md`, leaving `systemPrompt` out of the generated YAML.
 
 ## Memory
@@ -379,7 +379,7 @@ skills:
 
 For a private repository, store its access token in an AgentCore Identity API-key
 credential provider. The project deployment path resolves `auth.credentialName`
-from the [project credentials](../command.md#agentcore-project-add-credentials) declared in
+from the [project credentials](../command.md#agentcore-add-credentials) declared in
 `agentcore.json`:
 
 ```yaml
@@ -471,7 +471,7 @@ rather than being silently skipped.
 Exporting a Harness to a code-owned Runtime has additional limits: filesystem
 skills are rejected, and bundled AWS skills are omitted with an explanation in
 `EXPORT_NOTES.md`. S3 and Git sources are supported by the exporter. See
-[Export a Harness](../command.md#agentcore-project-export-harness).
+[Export a Harness](../command.md#agentcore-export-harness).
 
 See [Harness skills](https://docs.aws.amazon.com/bedrock-agentcore/latest/devguide/harness-skills.html)
 and the [Agent Skills format](https://agentskills.io/specification) for source
@@ -551,7 +551,7 @@ dockerfile: Dockerfile
 ```
 
 The path is relative to the directory containing `harness.yaml`. If supplied
-through `project add harness --dockerfile`, the CLI copies the file into the
+through `agentcore add harness --dockerfile`, the CLI copies the file into the
 harness directory as `Dockerfile`.
 
 Alternatively, reference a pre-built ECR image:

diff --git a/e2eTest/project/templates.test.ts b/e2eTest/project/templates.test.ts
@@ -161,16 +161,7 @@ describe(
       const created = parseResult(
         ProjectCreatedSchema,
         await cli.run(
-          [
-            "project",
-            "create",
-            "--name",
-            projectName,
-            "--template",
-            "empty",
-            "--skip-git",
-            "--json",
-          ],
+          ["create", "--name", projectName, "--template", "empty", "--skip-git", "--json"],
           projectRoot,
         ),
       );
@@ -184,16 +175,7 @@ describe(
         const added = parseResult(
           OperationSchema,
           await cli.run(
-            [
-              "project",
-              "add",
-              "runtime",
-              "--name",
-              runtime.name,
-              "--template",
-              runtime.template,
-              "--json",
-            ],
+            ["add", "runtime", "--name", runtime.name, "--template", runtime.template, "--json"],
             projectDir,
           ),
         );
@@ -220,7 +202,7 @@ describe(
       };
 
       beforeAll(() => {
-        dev = cli.start(["project", "dev", "--mode", "headless"], projectDir);
+        dev = cli.start(["dev", "--mode", "headless"], projectDir);
         dev.stdout?.on("data", captureDevOutput);
         dev.stderr?.on("data", captureDevOutput);
         dev.stdout?.resume();
@@ -242,11 +224,11 @@ describe(
           // the server may take a bit to get ready, so we retry on a timeout.
           const response = await retry(async () => {
             if (!dev) {
-              throw new Error(`project dev did not start. \nstdout/stdout = ${pendingOutput}`);
+              throw new Error(`agentcore dev did not start. \nstdout/stdout = ${pendingOutput}`);
             }
             if (dev.exitCode !== null) {
               throw new Error(
-                `project dev exited with code ${dev.exitCode ?? "unknown"}.  \nstdout/stdout = ${pendingOutput}`,
+                `agentcore dev exited with code ${dev.exitCode ?? "unknown"}.  \nstdout/stdout = ${pendingOutput}`,
               );
             }
 
@@ -261,7 +243,6 @@ describe(
               LocalRuntimeInvokeResponseSchema,
               await cli.run(
                 [
-                  "project",
                   "invoke",
                   "runtime",
                   "--local",
@@ -292,7 +273,7 @@ describe(
     test("deploys all runtimes", { timeout: TIMEOUT_MS.PROJECT_DEPLOY }, async () => {
       const deployment = parseResult(
         DeployResponseSchema,
-        await cli.run(["project", "deploy", "--yes", "--json"], projectDir),
+        await cli.run(["deploy", "--yes", "--json"], projectDir),
       );
       expect(deployment.message).toContain("Deployed project");
     });
@@ -306,7 +287,6 @@ describe(
           RuntimeInvokeResponseSchema,
           await cli.run(
             [
-              "project",
               "invoke",
               "runtime",
               "--name",
@@ -335,10 +315,7 @@ describe(
       async (runtime) => {
         const removed = parseResult(
           OperationSchema,
-          await cli.run(
-            ["project", "remove", "runtime", "--name", runtime.name, "--json"],
-            projectDir,
-          ),
+          await cli.run(["remove", "runtime", "--name", runtime.name, "--json"], projectDir),
         );
         expect(removed.operation).toBe("remove");
       },
@@ -350,13 +327,10 @@ describe(
       async () => {
         parseResult(
           JsonObjectSchema,
-          await cli.run(["project", "remove", "all", "--yes", "--json"], projectDir),
+          await cli.run(["remove", "all", "--yes", "--json"], projectDir),
         );
 
-        parseResult(
-          JsonObjectSchema,
-          await cli.run(["project", "deploy", "--yes", "--json"], projectDir),
-        );
+        parseResult(JsonObjectSchema, await cli.run(["deploy", "--yes", "--json"], projectDir));
       },
     );
   },

diff --git a/scripts/generate-command-reference.mjs b/scripts/generate-command-reference.mjs
@@ -9,7 +9,23 @@ const SCRIPT_DIR = dirname(fileURLToPath(import.meta.url));
 const REPOSITORY_ROOT = resolve(SCRIPT_DIR, "..");
 const DEFAULT_GROUPS = [
   { id: "global-options", title: "Global options", commands: [] },
-  { id: "project", title: "Project commands", commands: ["project"] },
+  {
+    id: "project",
+    title: "Project commands",
+    commands: [
+      "create",
+      "add",
+      "export",
+      "remove",
+      "dev",
+      "deploy",
+      "invoke",
+      "log",
+      "traces",
+      "status",
+      "build",
+    ],
+  },
   { id: "harness", title: "Harness commands", commands: ["harness"] },
   { id: "identity", title: "Identity commands", commands: ["identity"] },
   { id: "runtime", title: "Runtime commands", commands: ["runtime"] },

diff --git a/src/assets/cdk/README.md b/src/assets/cdk/README.md
@@ -17,9 +17,9 @@ the conventional `system-prompt.md` file in the harness directory.
 You normally do not run this app directly:
 
 ```bash
-agentcore project build    # synthesizes the CloudFormation templates into agentcore/cdk/cdk.out
-agentcore project deploy   # synthesizes, then deploys the stack for the selected target
-agentcore project status   # reports the resources agentcore.json declares
+agentcore build    # synthesizes the CloudFormation templates into agentcore/cdk/cdk.out
+agentcore deploy   # synthesizes, then deploys the stack for the selected target
+agentcore status   # reports the resources agentcore.json declares
 ```
 
 `npm run build` compiles the app, and `npx cdk synth` / `npx cdk diff` work from this directory too.
@@ -43,10 +43,10 @@ checkout.addEnvironmentVariable('ORDERS_TABLE', orders.tableName);
 orders.grantReadData(this.application.harness('support')); // any AWS L2 grant works too
 ```
 
-Then run `agentcore project deploy` again. An unknown name fails at synth and lists the names that exist.
+Then run `agentcore deploy` again. An unknown name fails at synth and lists the names that exist.
 
 If a runtime or harness is configured with an `executionRoleArn`, CDK cannot modify that imported role: every grant
 emits a synth-time warning listing the permissions that were not attached, and the role must already carry them.
 
-`agentcore project status` reports only the resources `agentcore.json` declares; resources you add here are visible
+`agentcore status` reports only the resources `agentcore.json` declares; resources you add here are visible
 through CloudFormation (`aws cloudformation describe-stack-resources`).
diff --git a/src/assets/evaluators/python-lambda/README.md b/src/assets/evaluators/python-lambda/README.md
@@ -28,5 +28,5 @@ def handler(input: EvaluatorInput, context) -> EvaluatorOutput:
     return EvaluatorOutput(value=1.0, label="Pass", explanation="…")
 ```
 
-Then `agentcore project deploy` packages this directory into the evaluator
+Then `agentcore deploy` packages this directory into the evaluator
 Lambda and registers the evaluator.
diff --git a/src/assets/templates/a2a-python-strands/README.md b/src/assets/templates/a2a-python-strands/README.md
@@ -21,11 +21,11 @@ def my_tool(param: str) -> str:
 
 ## Developing locally
 
-`agentcore project dev` starts the agent locally on `0.0.0.0:9000`. Fetch its agent card at
+`agentcore dev` starts the agent locally on `0.0.0.0:9000`. Fetch its agent card at
 `http://127.0.0.1:9000/.well-known/agent-card.json` and send it messages by posting A2A
 JSON-RPC to `http://127.0.0.1:9000/`.
 
 ## Deployment
 
-`agentcore project deploy` deploys the agent into Amazon Bedrock AgentCore. Invoke it with the
+`agentcore deploy` deploys the agent into Amazon Bedrock AgentCore. Invoke it with the
 AWS CLI (`bedrock-agentcore invoke-agent-runtime`) using an A2A JSON-RPC payload.
diff --git a/src/assets/templates/agent-python-langchain/README.md b/src/assets/templates/agent-python-langchain/README.md
@@ -13,16 +13,16 @@ session, and streams its response.
   between turns.
 - `model/load.py`: creates the Bedrock chat model with `init_chat_model`.
 - `pyproject.toml`: Python dependencies, managed with
-  [uv](https://docs.astral.sh/uv/). `agentcore project create` has already run
+  [uv](https://docs.astral.sh/uv/). `agentcore create` has already run
   `uv sync` for you (unless you passed `--skip-install`), so `.venv/` is ready.
 
 ## Develop
 
 Run the agent locally from the project root:
 
 ```bash
-agentcore project dev
-agentcore project invoke runtime --local --name {{name}} --payload '{"prompt":"What is 2 plus 3?"}'
+agentcore dev
+agentcore invoke runtime --local --name {{name}} --payload '{"prompt":"What is 2 plus 3?"}'
 ```
 
 Environment variables for local development go in `agentcore/.env.local`
@@ -31,8 +31,8 @@ Environment variables for local development go in `agentcore/.env.local`
 ## Deploy
 
 ```bash
-agentcore project deploy
-agentcore project invoke runtime --payload '{"prompt":"Hello!"}'
+agentcore deploy
+agentcore invoke runtime --payload '{"prompt":"Hello!"}'
 ```
 
 Traces are collected automatically: AgentCore Runtime starts the agent under

diff --git a/src/assets/templates/agent-python-minimal/README.md b/src/assets/templates/agent-python-minimal/README.md
@@ -9,15 +9,15 @@ invocation — a starting point you own and grow into a real agent.
 - `main.py` — the agent. A `BedrockAgentCoreApp` wraps the entrypoint that
   receives each invocation payload and returns the response.
 - `pyproject.toml` — Python dependencies, managed with
-  [uv](https://docs.astral.sh/uv/). `agentcore project create` has already run
+  [uv](https://docs.astral.sh/uv/). `agentcore create` has already run
   `uv sync` for you (unless you passed `--skip-install`), so `.venv/` is ready.
 
 ## Develop
 
 Run the agent locally from the project root:
 
 ```bash
-agentcore project dev
+agentcore dev
 ```
 
 Environment variables for local development go in `agentcore/.env.local`
@@ -26,6 +26,6 @@ Environment variables for local development go in `agentcore/.env.local`
 ## Deploy
 
 ```bash
-agentcore project deploy
-agentcore project invoke runtime --payload '{"prompt":"Hello!"}'
+agentcore deploy
+agentcore invoke runtime --payload '{"prompt":"Hello!"}'
 ```