Agent

Knowledge Base Management

AE Agent Platform Resource Management

Manage the Agent platform's models, MCP servers, Skills, automations, and attachment library through ae-cli, upgrading Agent configuration from manual web operations to precise command-line management.

IndustriesAll industries

Agent · Skills Live

skill · loadedindustry+100

Run this Skill on my data

Agent orchestration

multi-agent

Analysis readyevidence attached

The AE Agent Resource Management Skill is an Agent-platform resource-management capability in ThinkingAI's industry Skill library. It helps teams manage all Agent-platform resources through the ae-cli command-line tool, models (add/delete/enable-disable), MCP servers (add/delete/enable-disable), Skills (create/delete/enable-disable), automations (create scheduled/daily tasks), and the attachment library (upload/delete files), upgrading from configuring each item manually on the web to an efficient mode of precise batch management from the command line. This is ThinkingAI's decade of industry experience, packaged into ready-to-use Skills.

Live walkthrough

AgentSave as AgentExport session

ae-agent-resource-management

Set up the Agent environment for the new project: add a gpt-4o custom model, connect our internal data-query MCP, create a data-review Skill, and configure a daily-data-report automation that runs at 9 a.m. When it's done, give me a current resource inventory: I don't want to click through the web console item by item.

ae-cli agent +add-model to add the custom model gpt-4o (personal scope)

provider / base URL / API key validation passed

+add-mcp to connect the internal data-query service data-query-mcp

URL reachability pre-check passed (the platform doesn't verify connectivity at creation time)

+add-skill to create the 'Data Review' Skill: the long instruction block piped in via stdin

+create-automation to configure the 'Daily Data Report' automation, triggered daily at 09:00

Created, enabled by default

+add-attachment to upload the project config file project-config.yaml (214 KB)

Summarize the inventory of five resource types and their enabled status

The environment is set up. All five resource types are ready: 1 model, 1 MCP, 1 Skill, 1 daily automation, and 1 attachment, all from the command line in about 8 minutes, with no need to click through the web console item by item.

1. Current resource inventory

Resource	Name	Scope	Status
Model	gpt-4o (OpenAI provider)	personal	Enabled
MCP server	data-query-mcp	personal	Enabled · connectivity pre-check passed
Skill	Data Review	personal	Enabled
Skill	Instrumentation Validation	company	Enabled (read-only, can toggle)
Automation	Daily Data Report (daily 09:00)	personal	Enabled
Attachment	project-config.yaml (214 KB)	Attachment library	Uploaded · 0.9 GB quota remaining

← Scroll to see more

2. Notes on this run

The model and MCP are both created in personal scope: company / system-scope resources are read-only and can only be toggled, so this run won't affect anyone else.

Before adding the MCP, a URL reachability pre-check was run: the platform doesn't verify connectivity at creation time, so without the pre-check you wouldn't discover an unreachable service until runtime.

The 'Data Review' Skill's long instruction block was piped in via stdin, avoiding both the web text-box and the shell quote-escaping pitfalls.

The automation is enabled by default and runs for the first time at 09:00 tomorrow; to pause it, just update with --enabled false.

Recommendation

This initialization compressed from about 60 minutes of manual web work to 8 minutes. Save this run's command sequence (+add-model → +add-mcp → +add-skill → +create-automation → +add-attachment) as a project-init script, and the next project just changes the parameters and runs it in one go.

The 'Daily Data Report' automation is enabled and runs for the first time at 09:00 tomorrow.

On your data

That was a simulated run

Leave your work email and we will run a live walkthrough on your real business data.

The problem

The Agent platform is the core infrastructure of AI automation, but over 50% of teams are inefficient at Agent resource configuration. Adding a custom model requires manually entering the base URL, API key, provider, and other parameters (one wrong parameter and the connection fails); adding an MCP server requires manually configuring the URL and transport type (the platform does not verify connectivity, so an unreachable URL is only discovered at runtime); creating a Skill requires hand-writing long instruction text (awkward to do in a web text box); and the schedule config for automations (daily/weekly/custom time) must also be set one by one. An even more common problem is difficult resource discovery: no one knows which models, MCPs, and Skills are currently enabled, requiring a trip through the web every time, and enable/disable operations must be clicked one by one. Attachment management is a total gap: the reports, charts, and data files an Agent generates are scattered locally and cannot be uploaded in one place to the Agent's attachment library for later reference.

What it does

Full coverage of five resource types: models, MCP servers, Skills, automations, and attachment library, all managed in one place

Three-tier personal/company/system isolation: only personal-scope resources can be created and deleted, while company/system resources are read-only and can be enabled/disabled (affecting only the current user's preference)

Automation creation: supports daily/weekly/custom-time scheduled triggers, enabled by default with the option to pause

Skill instructions via stdin: long instruction text is piped in via stdin, avoiding shell quoting issues

Batch attachment upload: single files (up to 50MB) or batch files uploaded to the attachment library (1GB user quota), with partial failures not affecting other files

When to use it

Adding a custom model (OpenAI/Anthropic/other provider) to the Agent platform

Adding an MCP server to extend the Agent's tool capabilities

Creating a custom Skill to configure the Agent's behavior rules

Creating a scheduled automation (daily AI report, periodic data analysis, etc.)

Uploading files to the Agent attachment library for later reference

Viewing the list of currently enabled models/MCPs/Skills

In the field

Case

A data team · Agent environment setup for a new project

The team needed to configure a GPT-4o model, an internal MCP data-query server, a data-review Skill, a daily automated analysis task, and a project config file for a new project. With the AE Agent Resource Management Skill, the system ran +add-model, +add-mcp, +add-skill, +create-automation, and +add-attachment in sequence, cutting the configuration of 5 resource types from 60 minutes of manual web operations to 8 minutes of batch command-line execution.

FAQ

Can an MCP server be used right after it is added?

Not necessarily. The platform does not verify connectivity at creation, so a server with an unreachable URL is accepted but only errors at runtime. We recommend testing whether the URL is reachable after adding it.

Can company/system-scope resources be deleted?

No. The CLI only allows creating and deleting personal-scope resources. Company/system-scope models/MCPs/Skills can only be viewed and enabled/disabled (which affects only the current user's preference, not the global state).