Skip to content

TESTING_AND_CAPABILITIES.md

Latest commit

 

History

History
119 lines (82 loc) · 6.27 KB

TESTING_AND_CAPABILITIES.md

File metadata and controls

119 lines (82 loc) · 6.27 KB

AgentStack plugin testing and capabilities

Preparation for testing

Before manual testing, run the checklist and automatic validation:

  • VERIFICATION_CHECKLIST.md — step-by-step checklist: structure validation, install, MCP setup, chat scenarios.
  • Structure validation: from plugin root: node scripts/validate-plugin.mjs
  • MCP endpoint check (optional): .\scripts\test-mcp-endpoint.ps1 (PowerShell) or see "MCP check" block in the checklist.

How to verify the plugin works

1. Install the plugin

Option A: from Marketplace (after publish)

  • Cursor → Settings → Plugins (or Marketplace) → search "AgentStack" → Install.

Option B: locally (before publish)

  • Copy folder provided_plugins/cursor-plugin (or open the AgentStack project) to where Cursor loads plugins, or open the repo that contains the plugin.
  • Or add the plugin via Cursor UI if there is an "Add plugin from folder" option.

Current way to install a local plugin: Cursor Docs — Plugins.

2. MCP connection (required for calling tools)

The plugin provides Skills and Rules, but calls to projects, logic, buffs, etc. go through MCP. Without MCP configured, tools will not be called.

  1. Get an API key

    • Via curl (anonymous project): 
      curl -X POST https://agentstack.tech/mcp/tools/projects.create_project_anonymous \
  -H "Content-Type: application/json" \
  -d '{"tool": "projects.create_project_anonymous", "params": {"name": "Test"}}'
    • From the response take project_api_key or user_api_key.

  2. Add MCP server in Cursor

    • Settings → Features → Model Context Protocol (or MCP Servers).
    • Add Server: Name agentstack, Type HTTP, Base URL https://agentstack.tech/mcp.
    • In headers set X-API-Key = your key.

  3. Restart Cursor (if MCP did not appear).

Details: MCP_QUICKSTART.md.

3. Testing in chat

In Cursor chat ask the agent:

  • "Create a project in AgentStack named Test Project"
    → Expected: call to projects.create_project_anonymous (or projects.create_project when authenticated).
  • "Show my AgentStack projects"
    → Expected: projects.get_projects.
  • "Get stats for project <project_id>"
    → Expected: projects.get_stats.

If the agent calls MCP tools and returns a sensible answer — the plugin and MCP are working.

4. Checking Skills and Rules

  • Skills are picked up by Cursor and used when choosing "how to do" (8DNA, projects, Rules Engine). In logs/agent behavior you can confirm it follows Skill instructions (e.g. creates a project via MCP instead of writing its own HTTP client).
  • Rules (.mdc) apply to files by globs (e.g. when working with **/api/** or *.py). Code the agent suggests for AgentStack should follow the rules (data/config/protected structure, use of /api/*).

5. Common issues

Symptom What to check
Agent does not call MCP MCP added in Settings, correct Base URL and X-API-Key, Cursor restarted.
401 / 403 on call Key is valid, not expired; some operations require a subscription (e.g. Professional for add_user).
"Tool not found" Tool name matches documentation (e.g. projects.create_project_anonymous). Check list: GET https://agentstack.tech/mcp/tools (with X-API-Key header).
Skills not triggering Ensure plugin is installed and the skill description has the right trigger phrases (projects, 8DNA, rules, etc.).

Plugin capabilities

What the plugin includes

Component Purpose
Manifest (.cursor-plugin/plugin.json) Name, description, keywords for Marketplace and Cursor.
MCP config (mcp.json) Example AgentStack MCP server config (URL, API key header).
Skills (3) Teach the agent when and how to use AgentStack: 8DNA, projects, Rules Engine.
Rules (2 .mdc files) Code guidelines: data/config/protected structure and HTTP API usage (/api/*).
Documentation README, MCP_QUICKSTART, this file.

Capabilities via MCP (after MCP setup)

The plugin does not call the backend itself — the AgentStack MCP server does. After adding MCP in Cursor the agent gets access to tools such as:

  • Projects: create (including anonymous), list, details, update, delete, stats, users, settings, activity, API keys, attach anonymous project to user.
  • Logic and rules: create/update/delete rules, list, execute, processors, commands.
  • Buffs: create, apply, extend, rollback, cancel, list active, effective limits, temporary and persistent effects.
  • Payments: create, status, refund, list transactions, balance.
  • Auth: quick sign-in, create user, assign role, profile.
  • Scheduler: create/cancel/get/list tasks, etc.
  • Analytics: usage, metrics.
  • API keys: create, list, revoke, etc.
  • Webhooks, notifications, wallets — as implemented on backend and in MCP.

Full tool list and parameters: MCP_SERVER_CAPABILITIES in the AgentStack repo or GET https://agentstack.tech/mcp/tools (with X-API-Key).

Skills capabilities

  • agentstack-8dna: design and query data with hierarchy (parent_uuid) and evolution (generation), work with data/config/protected structure and genetic coding.
  • agentstack-projects: create and manage projects and API keys via MCP, anonymous projects, attach to user.
  • agentstack-rules-engine: configure server logic without code (when/do), use Logic Engine and rules via MCP, link with buffs and commands.

Rules capabilities (.mdc)

  • agentstack-dna-patterns: consistent patterns for data, config, protected structure and key naming when using AgentStack in code.
  • agentstack-json-config: when and how to use HTTP API (/api/projects, /api/logic, /api/neural, /api/buffs, etc.) and MCP.

Summary

  • Testing: install plugin → configure MCP with API key → in chat ask to create/list projects and verify MCP calls; optionally verify Skills and Rules from behavior and code.
  • Capabilities: access to 60+ AgentStack MCP tools (projects, logic, buffs, payments, auth, scheduler, analytics, etc.), plus three Skills and two Rules for consistent use of 8DNA, projects, and Rules Engine.