[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6996KP3GD7QBTD7J4GCF6NS":3,"skill-same:01M6996KP3GD7QBTD7J4GCF6NS":172},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":23,"publishedAt":43,"linkOnly":44,"markdownContent":45,"markdownBody":46,"frontMatter":47,"refName":51,"commitSha":52,"skillDirectory":53,"forksCount":54,"repositoryHtmlUrl":55,"repositoryHomepageUrl":6,"ownerHtmlUrl":56,"ownerBlogUrl":57,"ownerDescription":58,"updatedAt":59,"install":60,"resources":73},"01M6996KP3GD7QBTD7J4GCF6NS","google-cloud-solution-multi-agent-security",null,"Designs, deploys, and secures Google Cloud Agent Gateway solutions. Use when the user needs to configure multi-agent security, ingress (CLIENT_TO_AGENT), or egress (AGENT_TO_ANYWHERE) patterns involving Model Armor, IAP, and Agent Registry. Don't use for general Cloud Load Balancing or basic VPC setup not related to Agent Gateways.","google\u002Fskills","google","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F1342004?v=4","01M698VYEFDM0VJBZDYYJ2T83S","Google","01M698WR0H76ENHE587T9EJT89",24235,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002FSKILL.md",[18],{"categoryNo":19,"categoryName":20,"taxonomyType":21,"taxonomyTypeDesc":22},"01JBSEEDCAT000000000000023","Skill与智能体开发","SCENARIO","使用场景",[24,25,26,27,28,9,29,30,31,32,33,34,35,36,37,38,39,40,41,42],"agent security","multi agent","google cloud","security","google cloud solution multi agent security","multi","cloud","solutions","when","patterns","agent registry","multi-agent","model","configure","vpc","iap","gateways","gateway","registry","2026-11-10 10:12:30",false,"---\nname: google-cloud-solution-multi-agent-security\nmetadata:\n  version: \"1.0.1\"\n  category: MultiProductSolutions\ndescription: >-\n  Designs, deploys, and secures Google Cloud Agent Gateway solutions.\n  Use when the user needs to configure multi-agent security, ingress (CLIENT_TO_AGENT), or egress (AGENT_TO_ANYWHERE) patterns involving Model Armor, IAP, and Agent Registry.\n  Don't use for general Cloud Load Balancing or basic VPC setup not related to Agent Gateways.\n---\n\n# Agent Gateway multi-agent security\n\n## Attribution\n\n**MUST READ**: Before executing any `gcloud`, REST API, or MCP command, you\nMUST read and follow the instructions in\n[Usage Attribution](references\u002Fattribution.md).\n\n## Critical Enforcement Rules & Rationale\n\n*   **Gcloud Release Tracks**: Always use the exact release tracks specified in\n    the commands (e.g., `gcloud beta network-services agent-gateways`). Omitting\n    these prefixes causes commands to fail because Agent Gateway features are\n    located in specialized, non-default namespaces.\n*   **API Enablement**: Include `modelarmor.googleapis.com` in the API\n    enablement list when setting up guardrails. Excluding it prevents Model\n    Armor policies and filters from successfully attaching to the Gateway.\n*   **Egress Verification**: Egress policy verification requires using the\n    Python script\n    ([scripts\u002Fverify_egress_policies.py](scripts\u002Fverify_egress_policies.py)),\n    not `curl`. Egress gateways rely on runtime SDK lifecycle handling and JWT\n    context that a standard curl command cannot simulate correctly.\n*   **Model Armor Keys**: In `model-armor-config.yaml`, always include both\n    `piAndJailbreakFilterSettings` and `sdpFilterSettings` (`filterEnforcement:\n    ENFORCE`). Invalid or missing filters cause deployment validation failures\n    or lead to silent bypasses of the guardrails.\n*   **Subnet Private Access**: Any subnet hosting a Private Service Connect\n    network attachment for Egress Gateways must have `private_ip_google_access =\n    true` enabled in Terraform. Disabling this blocks connectivity to\n    Google-managed endpoints, causing total routing failures for agents.\n*   **Direct Delivery**: Immediately provide the requested architecture,\n    configuration files, CLI commands, scripts, and diagrams in full. Do not\n    stop at a planning phase, do not generate a plan artifact, and do not ask\n    for user confirmation before delivering outputs.\n*   **No Infrastructure Execution**: Do not attempt to run deployment or\n    verification commands (such as `gcloud`, `kubectl`, `terraform`, or `curl`)\n    against real cloud resources during design. You are generating plan\n    configurations, not executing them.\n\n> [!IMPORTANT] **Just-In-Time (JIT) Resource Loading Protocol:** Inspect\n> template files in [assets\u002F](assets\u002F) and executable scripts in\n> [scripts\u002F](scripts\u002F) using `view_file` as needed for extended configurations,\n> deployment scripts, and test suites.\n\n--------------------------------------------------------------------------------\n\n## Quick Reference: Required Filenames\n\nAlways generate files with these exact names when requested:\n\n1.  `agw-ingress-config.yaml`\n    ([assets\u002Fagw-ingress-config.yaml](assets\u002Fagw-ingress-config.yaml))\n2.  `agw-egress-config.yaml`\n    ([assets\u002Fagw-egress-config.yaml](assets\u002Fagw-egress-config.yaml))\n3.  `agw-authz-extension.yaml`\n    ([assets\u002Fagw-authz-extension.yaml](assets\u002Fagw-authz-extension.yaml))\n4.  `agw-authz-policy.yaml`\n    ([assets\u002Fagw-authz-policy.yaml](assets\u002Fagw-authz-policy.yaml))\n5.  `model-armor-config.yaml`\n    ([assets\u002Fmodel-armor-config.yaml](assets\u002Fmodel-armor-config.yaml))\n6.  `sgp-policy.yaml` ([assets\u002Fsgp-policy.yaml](assets\u002Fsgp-policy.yaml))\n7.  `iap-policy.json` ([assets\u002Fiap-policy.json](assets\u002Fiap-policy.json))\n8.  `model-armor-payload.json`\n    ([assets\u002Fmodel-armor-payload.json](assets\u002Fmodel-armor-payload.json))\n\n--------------------------------------------------------------------------------\n\n## 1. Dual Ingress & Egress Architecture Design (`dual_ingress_egress_architecture_design`)\n\n-   **Ingress Pattern**: `CLIENT_TO_AGENT` fronted by Ingress Control Plane\n    (Agent Gateway, Model Armor).\n-   **Egress Pattern**: `AGENT_TO_ANYWHERE` utilizing Egress Control Plane\n    (Agent Gateway, `roles\u002Fiap.egressor` CEL policies, Cloud DNS) and Egress\n    Data Plane (PSC Interface, Cloud Run, PSC Google APIs Global Endpoint),\n    coordinated via Agent Registry & Agent Engine runtime.\n-   **Mermaid Diagram**:\n\n    ```mermaid\n    graph TD\n        Client[\"External Clients\"] -->|HTTPS \u002F MCP| GLB[\"Global Load Balancer\"]\n        GLB --> Ingress[\"Ingress Agent Gateway (CLIENT_TO_AGENT)\"]\n        Ingress --> MA[\"Model Armor (CONTENT_AUTHZ)\"]\n        MA --> Agent[\"Agent Engine Agents (BillingAgent, SupportAgent, FraudAgent)\"]\n        Agent --> Egress[\"Egress Agent Gateway (AGENT_TO_ANYWHERE)\"]\n        Egress --> PSC[\"Private Service Connect Network Attachment\"]\n        PSC --> Tools[\"Private MCP Tool Backends\"]\n    ```\n\n--------------------------------------------------------------------------------\n\n## 2. Ingress & Egress Guardrail Policy Config (`ingress_and_egress_guardrail_policy_config`)\n\nWhen requested for Ingress & Egress guardrail policy configs, you MUST generate\nand create all required files in the workspace:\n\n-   `agw-ingress-config.yaml`\n    ([assets\u002Fagw-ingress-config.yaml](assets\u002Fagw-ingress-config.yaml)): Declares\n    `governedAccessPath: CLIENT_TO_AGENT` with protocols `HTTP` and `MCP`.\n-   `agw-egress-config.yaml`\n    ([assets\u002Fagw-egress-config.yaml](assets\u002Fagw-egress-config.yaml)): Declares\n    `governedAccessPath: AGENT_TO_ANYWHERE` with protocol `MCP`.\n-   `agw-authz-extension.yaml`\n    ([assets\u002Fagw-authz-extension.yaml](assets\u002Fagw-authz-extension.yaml)):\n    Configures AuthzExtension service for IAP authorization.\n-   `agw-authz-policy.yaml`\n    ([assets\u002Fagw-authz-policy.yaml](assets\u002Fagw-authz-policy.yaml)): Configures\n    `AuthzPolicy` action `ALLOW` targeting both Ingress and Egress gateways.\n-   `iap-policy.json` ([assets\u002Fiap-policy.json](assets\u002Fiap-policy.json)): Binds\n    `roles\u002Fiap.egressor` with CEL condition checking\n    `iap.googleapis.com\u002Fmcp.toolName == 'get_account_balance' &&\n    iap.googleapis.com\u002Fmcp.tool.isReadOnly == true`.\n-   `model-armor-config.yaml`\n    ([assets\u002Fmodel-armor-config.yaml](assets\u002Fmodel-armor-config.yaml)): Enables\n    `piAndJailbreakFilterSettings` and `sdpFilterSettings` with\n    `filterEnforcement: ENFORCE`.\n-   `sgp-policy.yaml` ([assets\u002Fsgp-policy.yaml](assets\u002Fsgp-policy.yaml)):\n    Implements Natural Language Constraints blocking transactions > $1000 and\n    sanitizing PII.\n\n--------------------------------------------------------------------------------\n\n## 3. Ingress & Egress Infrastructure Deployment (`ingress_and_egress_infrastructure_deployment`)\n\nInspect and provide the step-by-step `gcloud` CLI commands from\n[scripts\u002Fdeploy_infrastructure.sh](scripts\u002Fdeploy_infrastructure.sh):\n\n1.  **Enable Required APIs**: `compute`, `networkservices`, `networksecurity`,\n    `modelarmor`, `iap`, `agentregistry`, `serviceextensions`, and `aiplatform`.\n2.  **Import Agent Gateways**: Ingress (`agw-ingress-config.yaml`) and Egress\n    (`agw-egress-config.yaml`) via `gcloud alpha network-services agent-gateways\n    import`.\n3.  **Import Authz Extension**: `agw-authz-extension.yaml` via `gcloud beta\n    service-extensions authz-extensions import`.\n4.  **Import Authz Policy**: `agw-authz-policy.yaml` via `gcloud beta\n    network-security authz-policies import`.\n\n--------------------------------------------------------------------------------\n\n## 4. Ingress & Egress Security Validation (`ingress_and_egress_security_validation`)\n\nWhen validating security for Ingress and Egress:\n\n1.  **Ingress 403 Unauthenticated Test**: Provide the copy-pasteable\n    verification curl command from\n    [scripts\u002Fvalidate_ingress_unauth.sh](scripts\u002Fvalidate_ingress_unauth.sh)\n    sending an unauthenticated POST request to the Reasoning Engine endpoint\n    expecting HTTP 403 Forbidden.\n2.  **Python Egress Verification Script (MUST use Python script snippet, NOT\n    curl)**: Provide the Python verification script snippet from\n    [scripts\u002Fverify_egress_policies.py](scripts\u002Fverify_egress_policies.py)\n    sending JSON-RPC `tools\u002Fcall` requests (`get_account_balance`) through the\n    Egress Gateway to verify HTTP 200 for allowed tools.\n3.  **Model Armor Test Payload**: Generate `model-armor-payload.json`\n    ([assets\u002Fmodel-armor-payload.json](assets\u002Fmodel-armor-payload.json))\n    containing prompt injection\u002Fjailbreak instructions.\n\n--------------------------------------------------------------------------------\n\n## 5. Troubleshooting Ingress & Egress Failures (`troubleshooting_ingress_and_egress_failures`)\n\n-   **Ingress 403 (Client-to-Agent)**:\n\n    -   **Root Cause**: Unauthenticated client requests or missing\u002Finvalid OAuth\n        2.0 \u002F IAP identity tokens.\n    -   **OAuth Configuration Steps**:\n        1.  Configure OAuth 2.0 Client ID credentials in Google Cloud Console.\n        2.  Grant the client identity \u002F service account\n            `roles\u002Fiap.httpsResourceAccessor` permission.\n        3.  Exchange credentials with Google OAuth to acquire an OIDC \u002F OAuth ID\n            token.\n        4.  Pass the token in the `Authorization: Bearer \u003CTOKEN>` header.\n    -   **Verification Command**: Provide the curl command from\n        [scripts\u002Fverify_ingress_auth.sh](scripts\u002Fverify_ingress_auth.sh).\n\n-   **Egress 403 (Agent-to-Anywhere)**:\n\n    -   **Root Cause**: Missing `roles\u002Fiap.egressor` IAM bindings on the Agent\n        Identity, malformed principal ID, or mismatched CEL condition on tool\n        metadata.\n    -   **Fix Command**: Provide the exact `gcloud` command from\n        [scripts\u002Ffix_egress_iap.sh](scripts\u002Ffix_egress_iap.sh).\n\n--------------------------------------------------------------------------------\n\n## 6. Hybrid VPN Connectivity & Egress Routing (`hybrid_vpn_connectivity_egress_routing`)\n\n-   **Terraform HCL**: Refer to baseline Terraform config in\n    [assets\u002Fmain.tf](assets\u002Fmain.tf) for VPC, subnets\n    (`private_ip_google_access = true`), PSC network attachment, Cloud DNS\n    private forwarding for `aws.internal.`, and HA VPN gateway\u002Frouter.\n-   **Egress Gateway Config (`agw-egress-config.yaml`)**: Generate configuration\n    declaring `governedAccessPath: AGENT_TO_ANYWHERE`, pointing to the PSC\n    network attachment, and referencing `aws.internal.` in `dnsPeeringConfig`\n    (see [assets\u002Fagw-egress-config.yaml](assets\u002Fagw-egress-config.yaml)).\n-   **Python SDK Deployment Script**: Refer to\n    [scripts\u002Fhybrid_vpn_agent.py](scripts\u002Fhybrid_vpn_agent.py) for the complete\n    script initializing Vertex AI with `agent_to_anywhere_config` referencing\n    the Egress Gateway, enabling telemetry, and deploying `HybridAgent` using\n    `types.IdentityType.AGENT_IDENTITY`.\n\n--------------------------------------------------------------------------------\n\n## 7. Private Egress GKE Internal Load Balancer (`private_egress_gke_internal_load_balancer`)\n\n-   Expose GKE internal MCP tool server via an Internal Load Balancer (ILB) at\n    literal IP `10.0.1.50`, connecting via Agent Gateway PSC Interface + Cloud\n    DNS Private zone.\n-   **Cloud DNS Record Mapping**: Provide the command from\n    [scripts\u002Fcreate_gke_dns_record.sh](scripts\u002Fcreate_gke_dns_record.sh) mapping\n    the private domain to GKE's private ILB IP `10.0.1.50`.\n-   **Explicit TLS Warning**: Agent Gateway egress **does not natively trust\n    self-signed certificates or private enterprise CAs**. You **must** use\n    publicly trusted TLS certificates signed by a trusted Certificate Authority\n    (e.g., Let's Encrypt).\n\n--------------------------------------------------------------------------------\n\n## 8. Governance Controls Model Armor SGP (`governance_controls_model_armor_sgp`)\n\nWhen configuring dual safety layers with Model Armor on Ingress and SGP on\nEgress:\n\n1.  **Model Armor Config**: Generate `model-armor-config.yaml`\n    ([assets\u002Fmodel-armor-config.yaml](assets\u002Fmodel-armor-config.yaml)) with\n    `piAndJailbreakFilterSettings` and `sdpFilterSettings` (`filterEnforcement:\n    ENFORCE`).\n2.  **Semantic Governance Policy**: Generate `sgp-policy.yaml`\n    ([assets\u002Fsgp-policy.yaml](assets\u002Fsgp-policy.yaml)) with Natural Language\n    Constraints blocking transactions > $1000 and sanitizing PII.\n3.  **Curl PATCH Command**: Provide the curl command from\n    [scripts\u002Fenforce_sgp_patch.sh](scripts\u002Fenforce_sgp_patch.sh) to update\n    `authzExtensions` with `sgpEnforcementMode` set to `ENFORCE`.\n\n--------------------------------------------------------------------------------\n\n## 9. Multi-Agent Cloud Run Egress Routing (`multi_agent_cloud_run_egress_routing`)\n\nDo NOT produce a plan artifact or stop at planning. When configuring multi-agent\nCloud Run egress routing, you MUST directly provide and generate ALL required\ncomponents:\n\n1.  **Egress Gateway Config (`agw-egress-config-run.yaml`)**: Generate\n    configuration declaring `governedAccessPath: AGENT_TO_ANYWHERE`, PSC network\n    attachment, and DNS peering for `*.run.app` (see\n    [assets\u002Fagw-egress-config-run.yaml](assets\u002Fagw-egress-config-run.yaml)).\n2.  **Register Cloud Run Services in Agent Registry**: Provide the registration\n    commands from\n    [scripts\u002Fregister_cloud_run_services.sh](scripts\u002Fregister_cloud_run_services.sh)\n    registering all 3 Cloud Run services (`marketing-tool-service`,\n    `sales-tool-service`, `support-tool-service`) in the `us-east4` Agent\n    Registry.\n3.  **`iap-policy.json` (Multi-Agent)**: Generate `iap-policy.json`\n    ([assets\u002Fiap-policy-multi-agent.json](assets\u002Fiap-policy-multi-agent.json))\n    containing all 3 `principal:\u002F\u002F` bindings in the `members` list under\n    `roles\u002Fiap.egressor`.\n4.  **Python SDK Deployment Script**: Refer to\n    [scripts\u002Fmulti_agent_cloud_run.py](scripts\u002Fmulti_agent_cloud_run.py) for the\n    complete GenAI SDK deployment script.\n\n--------------------------------------------------------------------------------\n\n## 10. Advanced Model Armor Filtering (`advanced_model_armor_filtering`)\n\nFor custom keyword matching, configure `userDefinedFilterSettings` (see\n[assets\u002Fmodel-armor-advanced.yaml](assets\u002Fmodel-armor-advanced.yaml)).\n\n--------------------------------------------------------------------------------\n\n## 11. Known Traps & Gotchas (`known_traps_and_gotchas`)\n\n*   **`network_attachment` is `ForceNew`**: Enabling Semantic Governance\n    Policies (SGP) or modifying network attachments after the initial Terraform\n    apply will force-recreate the gateway resource. If not managed carefully,\n    this can cause dependency deadlocks during destroy operations. Plan\n    infrastructure sequencing accordingly.\n*   **Authz Policy Limit**: An Agent Gateway allows at most **4 custom\n    authorization policies** attached concurrently. Ensure your security posture\n    consolidates rules within this limit.\n","# Agent Gateway multi-agent security\n\n## Attribution\n\n**MUST READ**: Before executing any `gcloud`, REST API, or MCP command, you\nMUST read and follow the instructions in\n[Usage Attribution](references\u002Fattribution.md).\n\n## Critical Enforcement Rules & Rationale\n\n*   **Gcloud Release Tracks**: Always use the exact release tracks specified in\n    the commands (e.g., `gcloud beta network-services agent-gateways`). Omitting\n    these prefixes causes commands to fail because Agent Gateway features are\n    located in specialized, non-default namespaces.\n*   **API Enablement**: Include `modelarmor.googleapis.com` in the API\n    enablement list when setting up guardrails. Excluding it prevents Model\n    Armor policies and filters from successfully attaching to the Gateway.\n*   **Egress Verification**: Egress policy verification requires using the\n    Python script\n    ([scripts\u002Fverify_egress_policies.py](scripts\u002Fverify_egress_policies.py)),\n    not `curl`. Egress gateways rely on runtime SDK lifecycle handling and JWT\n    context that a standard curl command cannot simulate correctly.\n*   **Model Armor Keys**: In `model-armor-config.yaml`, always include both\n    `piAndJailbreakFilterSettings` and `sdpFilterSettings` (`filterEnforcement:\n    ENFORCE`). Invalid or missing filters cause deployment validation failures\n    or lead to silent bypasses of the guardrails.\n*   **Subnet Private Access**: Any subnet hosting a Private Service Connect\n    network attachment for Egress Gateways must have `private_ip_google_access =\n    true` enabled in Terraform. Disabling this blocks connectivity to\n    Google-managed endpoints, causing total routing failures for agents.\n*   **Direct Delivery**: Immediately provide the requested architecture,\n    configuration files, CLI commands, scripts, and diagrams in full. Do not\n    stop at a planning phase, do not generate a plan artifact, and do not ask\n    for user confirmation before delivering outputs.\n*   **No Infrastructure Execution**: Do not attempt to run deployment or\n    verification commands (such as `gcloud`, `kubectl`, `terraform`, or `curl`)\n    against real cloud resources during design. You are generating plan\n    configurations, not executing them.\n\n> [!IMPORTANT] **Just-In-Time (JIT) Resource Loading Protocol:** Inspect\n> template files in [assets\u002F](assets\u002F) and executable scripts in\n> [scripts\u002F](scripts\u002F) using `view_file` as needed for extended configurations,\n> deployment scripts, and test suites.\n\n--------------------------------------------------------------------------------\n\n## Quick Reference: Required Filenames\n\nAlways generate files with these exact names when requested:\n\n1.  `agw-ingress-config.yaml`\n    ([assets\u002Fagw-ingress-config.yaml](assets\u002Fagw-ingress-config.yaml))\n2.  `agw-egress-config.yaml`\n    ([assets\u002Fagw-egress-config.yaml](assets\u002Fagw-egress-config.yaml))\n3.  `agw-authz-extension.yaml`\n    ([assets\u002Fagw-authz-extension.yaml](assets\u002Fagw-authz-extension.yaml))\n4.  `agw-authz-policy.yaml`\n    ([assets\u002Fagw-authz-policy.yaml](assets\u002Fagw-authz-policy.yaml))\n5.  `model-armor-config.yaml`\n    ([assets\u002Fmodel-armor-config.yaml](assets\u002Fmodel-armor-config.yaml))\n6.  `sgp-policy.yaml` ([assets\u002Fsgp-policy.yaml](assets\u002Fsgp-policy.yaml))\n7.  `iap-policy.json` ([assets\u002Fiap-policy.json](assets\u002Fiap-policy.json))\n8.  `model-armor-payload.json`\n    ([assets\u002Fmodel-armor-payload.json](assets\u002Fmodel-armor-payload.json))\n\n--------------------------------------------------------------------------------\n\n## 1. Dual Ingress & Egress Architecture Design (`dual_ingress_egress_architecture_design`)\n\n-   **Ingress Pattern**: `CLIENT_TO_AGENT` fronted by Ingress Control Plane\n    (Agent Gateway, Model Armor).\n-   **Egress Pattern**: `AGENT_TO_ANYWHERE` utilizing Egress Control Plane\n    (Agent Gateway, `roles\u002Fiap.egressor` CEL policies, Cloud DNS) and Egress\n    Data Plane (PSC Interface, Cloud Run, PSC Google APIs Global Endpoint),\n    coordinated via Agent Registry & Agent Engine runtime.\n-   **Mermaid Diagram**:\n\n    ```mermaid\n    graph TD\n        Client[\"External Clients\"] -->|HTTPS \u002F MCP| GLB[\"Global Load Balancer\"]\n        GLB --> Ingress[\"Ingress Agent Gateway (CLIENT_TO_AGENT)\"]\n        Ingress --> MA[\"Model Armor (CONTENT_AUTHZ)\"]\n        MA --> Agent[\"Agent Engine Agents (BillingAgent, SupportAgent, FraudAgent)\"]\n        Agent --> Egress[\"Egress Agent Gateway (AGENT_TO_ANYWHERE)\"]\n        Egress --> PSC[\"Private Service Connect Network Attachment\"]\n        PSC --> Tools[\"Private MCP Tool Backends\"]\n    ```\n\n--------------------------------------------------------------------------------\n\n## 2. Ingress & Egress Guardrail Policy Config (`ingress_and_egress_guardrail_policy_config`)\n\nWhen requested for Ingress & Egress guardrail policy configs, you MUST generate\nand create all required files in the workspace:\n\n-   `agw-ingress-config.yaml`\n    ([assets\u002Fagw-ingress-config.yaml](assets\u002Fagw-ingress-config.yaml)): Declares\n    `governedAccessPath: CLIENT_TO_AGENT` with protocols `HTTP` and `MCP`.\n-   `agw-egress-config.yaml`\n    ([assets\u002Fagw-egress-config.yaml](assets\u002Fagw-egress-config.yaml)): Declares\n    `governedAccessPath: AGENT_TO_ANYWHERE` with protocol `MCP`.\n-   `agw-authz-extension.yaml`\n    ([assets\u002Fagw-authz-extension.yaml](assets\u002Fagw-authz-extension.yaml)):\n    Configures AuthzExtension service for IAP authorization.\n-   `agw-authz-policy.yaml`\n    ([assets\u002Fagw-authz-policy.yaml](assets\u002Fagw-authz-policy.yaml)): Configures\n    `AuthzPolicy` action `ALLOW` targeting both Ingress and Egress gateways.\n-   `iap-policy.json` ([assets\u002Fiap-policy.json](assets\u002Fiap-policy.json)): Binds\n    `roles\u002Fiap.egressor` with CEL condition checking\n    `iap.googleapis.com\u002Fmcp.toolName == 'get_account_balance' &&\n    iap.googleapis.com\u002Fmcp.tool.isReadOnly == true`.\n-   `model-armor-config.yaml`\n    ([assets\u002Fmodel-armor-config.yaml](assets\u002Fmodel-armor-config.yaml)): Enables\n    `piAndJailbreakFilterSettings` and `sdpFilterSettings` with\n    `filterEnforcement: ENFORCE`.\n-   `sgp-policy.yaml` ([assets\u002Fsgp-policy.yaml](assets\u002Fsgp-policy.yaml)):\n    Implements Natural Language Constraints blocking transactions > $1000 and\n    sanitizing PII.\n\n--------------------------------------------------------------------------------\n\n## 3. Ingress & Egress Infrastructure Deployment (`ingress_and_egress_infrastructure_deployment`)\n\nInspect and provide the step-by-step `gcloud` CLI commands from\n[scripts\u002Fdeploy_infrastructure.sh](scripts\u002Fdeploy_infrastructure.sh):\n\n1.  **Enable Required APIs**: `compute`, `networkservices`, `networksecurity`,\n    `modelarmor`, `iap`, `agentregistry`, `serviceextensions`, and `aiplatform`.\n2.  **Import Agent Gateways**: Ingress (`agw-ingress-config.yaml`) and Egress\n    (`agw-egress-config.yaml`) via `gcloud alpha network-services agent-gateways\n    import`.\n3.  **Import Authz Extension**: `agw-authz-extension.yaml` via `gcloud beta\n    service-extensions authz-extensions import`.\n4.  **Import Authz Policy**: `agw-authz-policy.yaml` via `gcloud beta\n    network-security authz-policies import`.\n\n--------------------------------------------------------------------------------\n\n## 4. Ingress & Egress Security Validation (`ingress_and_egress_security_validation`)\n\nWhen validating security for Ingress and Egress:\n\n1.  **Ingress 403 Unauthenticated Test**: Provide the copy-pasteable\n    verification curl command from\n    [scripts\u002Fvalidate_ingress_unauth.sh](scripts\u002Fvalidate_ingress_unauth.sh)\n    sending an unauthenticated POST request to the Reasoning Engine endpoint\n    expecting HTTP 403 Forbidden.\n2.  **Python Egress Verification Script (MUST use Python script snippet, NOT\n    curl)**: Provide the Python verification script snippet from\n    [scripts\u002Fverify_egress_policies.py](scripts\u002Fverify_egress_policies.py)\n    sending JSON-RPC `tools\u002Fcall` requests (`get_account_balance`) through the\n    Egress Gateway to verify HTTP 200 for allowed tools.\n3.  **Model Armor Test Payload**: Generate `model-armor-payload.json`\n    ([assets\u002Fmodel-armor-payload.json](assets\u002Fmodel-armor-payload.json))\n    containing prompt injection\u002Fjailbreak instructions.\n\n--------------------------------------------------------------------------------\n\n## 5. Troubleshooting Ingress & Egress Failures (`troubleshooting_ingress_and_egress_failures`)\n\n-   **Ingress 403 (Client-to-Agent)**:\n\n    -   **Root Cause**: Unauthenticated client requests or missing\u002Finvalid OAuth\n        2.0 \u002F IAP identity tokens.\n    -   **OAuth Configuration Steps**:\n        1.  Configure OAuth 2.0 Client ID credentials in Google Cloud Console.\n        2.  Grant the client identity \u002F service account\n            `roles\u002Fiap.httpsResourceAccessor` permission.\n        3.  Exchange credentials with Google OAuth to acquire an OIDC \u002F OAuth ID\n            token.\n        4.  Pass the token in the `Authorization: Bearer \u003CTOKEN>` header.\n    -   **Verification Command**: Provide the curl command from\n        [scripts\u002Fverify_ingress_auth.sh](scripts\u002Fverify_ingress_auth.sh).\n\n-   **Egress 403 (Agent-to-Anywhere)**:\n\n    -   **Root Cause**: Missing `roles\u002Fiap.egressor` IAM bindings on the Agent\n        Identity, malformed principal ID, or mismatched CEL condition on tool\n        metadata.\n    -   **Fix Command**: Provide the exact `gcloud` command from\n        [scripts\u002Ffix_egress_iap.sh](scripts\u002Ffix_egress_iap.sh).\n\n--------------------------------------------------------------------------------\n\n## 6. Hybrid VPN Connectivity & Egress Routing (`hybrid_vpn_connectivity_egress_routing`)\n\n-   **Terraform HCL**: Refer to baseline Terraform config in\n    [assets\u002Fmain.tf](assets\u002Fmain.tf) for VPC, subnets\n    (`private_ip_google_access = true`), PSC network attachment, Cloud DNS\n    private forwarding for `aws.internal.`, and HA VPN gateway\u002Frouter.\n-   **Egress Gateway Config (`agw-egress-config.yaml`)**: Generate configuration\n    declaring `governedAccessPath: AGENT_TO_ANYWHERE`, pointing to the PSC\n    network attachment, and referencing `aws.internal.` in `dnsPeeringConfig`\n    (see [assets\u002Fagw-egress-config.yaml](assets\u002Fagw-egress-config.yaml)).\n-   **Python SDK Deployment Script**: Refer to\n    [scripts\u002Fhybrid_vpn_agent.py](scripts\u002Fhybrid_vpn_agent.py) for the complete\n    script initializing Vertex AI with `agent_to_anywhere_config` referencing\n    the Egress Gateway, enabling telemetry, and deploying `HybridAgent` using\n    `types.IdentityType.AGENT_IDENTITY`.\n\n--------------------------------------------------------------------------------\n\n## 7. Private Egress GKE Internal Load Balancer (`private_egress_gke_internal_load_balancer`)\n\n-   Expose GKE internal MCP tool server via an Internal Load Balancer (ILB) at\n    literal IP `10.0.1.50`, connecting via Agent Gateway PSC Interface + Cloud\n    DNS Private zone.\n-   **Cloud DNS Record Mapping**: Provide the command from\n    [scripts\u002Fcreate_gke_dns_record.sh](scripts\u002Fcreate_gke_dns_record.sh) mapping\n    the private domain to GKE's private ILB IP `10.0.1.50`.\n-   **Explicit TLS Warning**: Agent Gateway egress **does not natively trust\n    self-signed certificates or private enterprise CAs**. You **must** use\n    publicly trusted TLS certificates signed by a trusted Certificate Authority\n    (e.g., Let's Encrypt).\n\n--------------------------------------------------------------------------------\n\n## 8. Governance Controls Model Armor SGP (`governance_controls_model_armor_sgp`)\n\nWhen configuring dual safety layers with Model Armor on Ingress and SGP on\nEgress:\n\n1.  **Model Armor Config**: Generate `model-armor-config.yaml`\n    ([assets\u002Fmodel-armor-config.yaml](assets\u002Fmodel-armor-config.yaml)) with\n    `piAndJailbreakFilterSettings` and `sdpFilterSettings` (`filterEnforcement:\n    ENFORCE`).\n2.  **Semantic Governance Policy**: Generate `sgp-policy.yaml`\n    ([assets\u002Fsgp-policy.yaml](assets\u002Fsgp-policy.yaml)) with Natural Language\n    Constraints blocking transactions > $1000 and sanitizing PII.\n3.  **Curl PATCH Command**: Provide the curl command from\n    [scripts\u002Fenforce_sgp_patch.sh](scripts\u002Fenforce_sgp_patch.sh) to update\n    `authzExtensions` with `sgpEnforcementMode` set to `ENFORCE`.\n\n--------------------------------------------------------------------------------\n\n## 9. Multi-Agent Cloud Run Egress Routing (`multi_agent_cloud_run_egress_routing`)\n\nDo NOT produce a plan artifact or stop at planning. When configuring multi-agent\nCloud Run egress routing, you MUST directly provide and generate ALL required\ncomponents:\n\n1.  **Egress Gateway Config (`agw-egress-config-run.yaml`)**: Generate\n    configuration declaring `governedAccessPath: AGENT_TO_ANYWHERE`, PSC network\n    attachment, and DNS peering for `*.run.app` (see\n    [assets\u002Fagw-egress-config-run.yaml](assets\u002Fagw-egress-config-run.yaml)).\n2.  **Register Cloud Run Services in Agent Registry**: Provide the registration\n    commands from\n    [scripts\u002Fregister_cloud_run_services.sh](scripts\u002Fregister_cloud_run_services.sh)\n    registering all 3 Cloud Run services (`marketing-tool-service`,\n    `sales-tool-service`, `support-tool-service`) in the `us-east4` Agent\n    Registry.\n3.  **`iap-policy.json` (Multi-Agent)**: Generate `iap-policy.json`\n    ([assets\u002Fiap-policy-multi-agent.json](assets\u002Fiap-policy-multi-agent.json))\n    containing all 3 `principal:\u002F\u002F` bindings in the `members` list under\n    `roles\u002Fiap.egressor`.\n4.  **Python SDK Deployment Script**: Refer to\n    [scripts\u002Fmulti_agent_cloud_run.py](scripts\u002Fmulti_agent_cloud_run.py) for the\n    complete GenAI SDK deployment script.\n\n--------------------------------------------------------------------------------\n\n## 10. Advanced Model Armor Filtering (`advanced_model_armor_filtering`)\n\nFor custom keyword matching, configure `userDefinedFilterSettings` (see\n[assets\u002Fmodel-armor-advanced.yaml](assets\u002Fmodel-armor-advanced.yaml)).\n\n--------------------------------------------------------------------------------\n\n## 11. Known Traps & Gotchas (`known_traps_and_gotchas`)\n\n*   **`network_attachment` is `ForceNew`**: Enabling Semantic Governance\n    Policies (SGP) or modifying network attachments after the initial Terraform\n    apply will force-recreate the gateway resource. If not managed carefully,\n    this can cause dependency deadlocks during destroy operations. Plan\n    infrastructure sequencing accordingly.\n*   **Authz Policy Limit**: An Agent Gateway allows at most **4 custom\n    authorization policies** attached concurrently. Ensure your security posture\n    consolidates rules within this limit.\n",{"name":5,"metadata":48,"description":7},{"version":49,"category":50},"1.0.1","MultiProductSolutions","main","bb9b3f464aae37af360d079c1d863524955b0363","skills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security",2014,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills","https:\u002F\u002Fgithub.com\u002Fgoogle","https:\u002F\u002Fopensource.google\u002F","Google ❤️ Open Source","2026-11-14 23:31:00",{"source":8,"skillName":5,"commands":61,"prompt":71,"zipAvailable":72},[62,65,68],{"tool":63,"command":64},"npx","npx skills add google\u002Fskills --skill google-cloud-solution-multi-agent-security",{"tool":66,"command":67},"bunx","bunx skills add google\u002Fskills --skill google-cloud-solution-multi-agent-security",{"tool":69,"command":70},"pnpm","pnpm dlx skills add google\u002Fskills --skill google-cloud-solution-multi-agent-security","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002FSKILL.md",true,[74,80,84,88,92,96,100,104,108,112,116,120,124,130,136,140,144,148,152,156,160,164,168],{"path":75,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":78,"sourceHtmlUrl":79},"assets\u002Fagw-authz-extension.yaml","ASSET","资产文件",275,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002Fassets\u002Fagw-authz-extension.yaml",{"path":81,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":82,"sourceHtmlUrl":83},"assets\u002Fagw-authz-policy.yaml",325,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002Fassets\u002Fagw-authz-policy.yaml",{"path":85,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":86,"sourceHtmlUrl":87},"assets\u002Fagw-egress-config-run.yaml",530,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002Fassets\u002Fagw-egress-config-run.yaml",{"path":89,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":90,"sourceHtmlUrl":91},"assets\u002Fagw-egress-config.yaml",247,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002Fassets\u002Fagw-egress-config.yaml",{"path":93,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":94,"sourceHtmlUrl":95},"assets\u002Fagw-ingress-config.yaml",255,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002Fassets\u002Fagw-ingress-config.yaml",{"path":97,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":98,"sourceHtmlUrl":99},"assets\u002Fiap-policy-multi-agent.json",590,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002Fassets\u002Fiap-policy-multi-agent.json",{"path":101,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":102,"sourceHtmlUrl":103},"assets\u002Fiap-policy.json",469,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002Fassets\u002Fiap-policy.json",{"path":105,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":106,"sourceHtmlUrl":107},"assets\u002Fmain.tf",2421,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002Fassets\u002Fmain.tf",{"path":109,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":110,"sourceHtmlUrl":111},"assets\u002Fmodel-armor-advanced.yaml",386,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002Fassets\u002Fmodel-armor-advanced.yaml",{"path":113,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":114,"sourceHtmlUrl":115},"assets\u002Fmodel-armor-config.yaml",353,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002Fassets\u002Fmodel-armor-config.yaml",{"path":117,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":118,"sourceHtmlUrl":119},"assets\u002Fmodel-armor-payload.json",101,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002Fassets\u002Fmodel-armor-payload.json",{"path":121,"resourceType":76,"resourceTypeDesc":77,"sizeBytes":122,"sourceHtmlUrl":123},"assets\u002Fsgp-policy.yaml",311,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002Fassets\u002Fsgp-policy.yaml",{"path":125,"resourceType":126,"resourceTypeDesc":127,"sizeBytes":128,"sourceHtmlUrl":129},"references\u002Fattribution.md","REFERENCE","参考文档",1974,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002Freferences\u002Fattribution.md",{"path":131,"resourceType":132,"resourceTypeDesc":133,"sizeBytes":134,"sourceHtmlUrl":135},"scripts\u002Fcreate_gke_dns_record.sh","SCRIPT","脚本",287,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002Fscripts\u002Fcreate_gke_dns_record.sh",{"path":137,"resourceType":132,"resourceTypeDesc":133,"sizeBytes":138,"sourceHtmlUrl":139},"scripts\u002Fdeploy_infrastructure.sh",1052,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002Fscripts\u002Fdeploy_infrastructure.sh",{"path":141,"resourceType":132,"resourceTypeDesc":133,"sizeBytes":142,"sourceHtmlUrl":143},"scripts\u002Fenforce_sgp_patch.sh",548,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002Fscripts\u002Fenforce_sgp_patch.sh",{"path":145,"resourceType":132,"resourceTypeDesc":133,"sizeBytes":146,"sourceHtmlUrl":147},"scripts\u002Ffix_egress_iap.sh",226,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002Fscripts\u002Ffix_egress_iap.sh",{"path":149,"resourceType":132,"resourceTypeDesc":133,"sizeBytes":150,"sourceHtmlUrl":151},"scripts\u002Fhybrid_vpn_agent.py",1216,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002Fscripts\u002Fhybrid_vpn_agent.py",{"path":153,"resourceType":132,"resourceTypeDesc":133,"sizeBytes":154,"sourceHtmlUrl":155},"scripts\u002Fmulti_agent_cloud_run.py",1760,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002Fscripts\u002Fmulti_agent_cloud_run.py",{"path":157,"resourceType":132,"resourceTypeDesc":133,"sizeBytes":158,"sourceHtmlUrl":159},"scripts\u002Fregister_cloud_run_services.sh",1065,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002Fscripts\u002Fregister_cloud_run_services.sh",{"path":161,"resourceType":132,"resourceTypeDesc":133,"sizeBytes":162,"sourceHtmlUrl":163},"scripts\u002Fvalidate_ingress_unauth.sh",433,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002Fscripts\u002Fvalidate_ingress_unauth.sh",{"path":165,"resourceType":132,"resourceTypeDesc":133,"sizeBytes":166,"sourceHtmlUrl":167},"scripts\u002Fverify_egress_policies.py",797,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002Fscripts\u002Fverify_egress_policies.py",{"path":169,"resourceType":132,"resourceTypeDesc":133,"sizeBytes":170,"sourceHtmlUrl":171},"scripts\u002Fverify_ingress_auth.sh",483,"https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fgoogle-cloud-solution-multi-agent-security\u002Fscripts\u002Fverify_ingress_auth.sh",[173,181,188,195,202,209],{"skillNo":174,"name":175,"nameZh":6,"description":176,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":177,"categories":178,"keywords":179,"publishedAt":180},"01M7DQ9K7JF4T7NJZ2K1MQZ44Q","cloud-operator-share-context","Shares Google Cloud operating context the user or their team already have — environments, projects, services and regions, product preferences, best practices, architecture guidance, runbooks — found in agent instruction files, local memory, skills and workflows, and the repository's GCP docs, with the operator agent's memory after the user reviews the list. Use when the user runs this skill, accepts the cloud-operator or cloud-operator-setup skill's offer to share their existing Google Cloud context, or asks to share, import or save that context or a GCP fact for the operator agent. Don't use for ordinary GCP requests, which go to the cloud-operator skill.","https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fcloud-operator-share-context\u002FSKILL.md",[],[],"2026-11-15 14:19:17",{"skillNo":182,"name":183,"nameZh":6,"description":184,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":185,"categories":186,"keywords":187,"publishedAt":180},"01M7DQ9ERFGGQCFAWQW0KRKYBY","cloud-operator-setup","Sets up the operator agent in a Google Cloud project via the autocloud MCP server: picks the operator project, creates or checks the user's operator agent, saves the project, and offers to share the user's existing Google Cloud operating context with the operator agent's memory. Use when the user runs this skill, asks to set up the operator or create their operator agent, or chooses setup when the cloud-operator skill asks. Don't use for ordinary GCP requests or installing gcloud.","https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fcloud-operator-setup\u002FSKILL.md",[],[],{"skillNo":189,"name":190,"nameZh":6,"description":191,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":192,"categories":193,"keywords":194,"publishedAt":180},"01M7B4RHQCM0DK2XQ7S7ZXVN8D","developer-device-platform-device-run","Runs automated, non-interactive jobs on remote Android devices with Developer Device Platform (DDP) Device Run using `gcloud beta device-run sessions`: native Android executables or shell scripts (for example CLI tools, custom runners, test harnesses or benchmarks) and Android instrumentation tests, on one or many devices, then waits for completion and collects per-device outputs and logs. Use when running a binary or test suite on remote devices without an interactive connection. Don't use it for interactive device reservations, adb tunnels or screen streaming (use developer-device-platform-basics), or for local devices.","https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fdeveloper-device-platform-device-run\u002FSKILL.md",[],[],{"skillNo":196,"name":197,"nameZh":6,"description":198,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":199,"categories":200,"keywords":201,"publishedAt":180},"01M7B4R9YXMRW7RZP9Q9FMZJ2P","compute-vm-creation","Guides the planning, configuration, and provisioning of standalone Google Compute Engine virtual machines (VMs) and bare metal (-metal) instances using gcloud CLI commands or direct MCP tool execution. Use ONLY when the user wants to create, configure, or provision a new standalone Compute Engine VM or bare metal instance from conversational intents (hardware sizing, region selection, OS image, networking, security defaults). DO NOT read or use for Google Kubernetes Engine (GKE) Autopilot or Standard clusters, Kubernetes node pools or Pods, container orchestration, managing existing VMs, Managed Instance Groups (MIGs), or Instance Templates.","https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fcompute-vm-creation\u002FSKILL.md",[],[],{"skillNo":203,"name":204,"nameZh":6,"description":205,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":206,"categories":207,"keywords":208,"publishedAt":180},"01M7B4QD02N0YBVT0838Z5K5VC","agent-platform-autorater-customization","Customizes an autorater (evaluation metric) for an AI agent so its grades match the user's own judgment, starting from the agent's real trajectories. Use when a user wants a metric tailored to their agent, wants to turn failing trajectories into rubrics, says an existing autorater grades their agent wrong, or wants an autorater to agree with human labels (autorater hillclimbing, judge calibration). Don't use for quick evals with built-in metrics or for improving the agent itself (use agent-platform-eval-flywheel).","https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fagent-platform-autorater-customization\u002FSKILL.md",[],[],{"skillNo":210,"name":211,"nameZh":6,"description":212,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":213,"categories":214,"keywords":215,"publishedAt":180},"01M78JP22K73KK96JH0JNDT5C4","iap-agent-platform-troubleshooting","Diagnoses Identity-Aware Proxy (IAP) denials of outbound requests from Gemini Enterprise Agent Platform agents that run on Agent Runtime and send traffic through Agent Gateway. Use when an agent request fails with \"Egress request is not authorized\", when Cloud Audit Logs show an IAP egressViaIAP decision with granted false, when IAP audit entries are missing while debugging agent egress, or when Access policies, IAP allow policies, policy bindings, or Agent Registry entries might block agent egress. Don't use for Gemini model IAM permission errors (such as aiplatform.endpoints.predict), Agent Platform API errors, or Agent Runtime deployment or 500 errors (use agent-platform-troubleshooting instead). Don't use for setting up, creating, or connecting an Agent Gateway, or for IAP web apps, IAP TCP forwarding, or SSH.","https:\u002F\u002Fgithub.com\u002Fgoogle\u002Fskills\u002Fblob\u002Fmain\u002Fskills\u002Fcloud\u002Fiap-agent-platform-troubleshooting\u002FSKILL.md",[],[]]