pitcrew-agents/pitcrew
v0.1.0MIT
Build and run AI compliance agents by chatting. Describe an agent in plain English and PitCrew drafts it, publishes it to your workspace, and runs it on your documents, without leaving the conversation.
hosts
- mcp.gopitcrew.com
plugin.json
- name
- pitcrew
- version
- 0.1.0
- description
- Build and run AI compliance agents by chatting. Describe an agent in plain English and PitCrew drafts it, publishes it to your workspace, and runs it on your documents, without leaving the conversation.
- author
- PitCrewhttps://www.gopitcrew.com/
- license
- MIT
- homepage
- https://www.gopitcrew.com/
- extensions
- com.openai
mcp.json
- PitCrew MCPstreamable-httphttps://mcp.gopitcrew.com/mcp
{
"$schema": "https://agent-plugins.org/schemas/1.0.0/plugin.schema.json",
"name": "pitcrew",
"version": "0.1.0",
"description": "Build and run AI compliance agents by chatting. Describe an agent in plain English and PitCrew drafts it, publishes it to your workspace, and runs it on your documents, without leaving the conversation.",
"author": {
"name": "PitCrew",
"url": "https://www.gopitcrew.com/"
},
"homepage": "https://www.gopitcrew.com/",
"repository": "https://github.com/pitcrew-agents/pitcrew-plugin",
"license": "MIT",
"keywords": [
"compliance",
"ai-agents",
"workflow-automation",
"financial-services",
"advisors",
"document-processing"
],
"extensions": {
"com.openai": {
"interface": {
"displayName": "PitCrew",
"shortDescription": "Build AI compliance agents",
"longDescription": "PitCrew is an AI agent platform for the work that teams repeat every day. The PitCrew connector lets you start, check and create agents and assets without leaving ChatGPT.\n\nNow you can:\n- Create a proposal: \"Draft a proposal for the Johnson account from yesterday's meeting notes.\"\n- Onboard a client: \"Set up the Miller account and list what is still missing from their file.\"\n- Fill forms: \"Complete the intake forms for the Anderson engagement from the documents we have.\"\n- Prepare for a meeting: \"Build the brief for Thursday's review with Thompson.\"\n- Check billing: \"Validate September invoices against the contracts and flag anything over policy.\"\n- Reconcile: \"Compare the two system exports for Q3 and show the breaks.\"\n- Follow up: \"Draft this week's post-meeting follow-ups for approval.\"\n- Check status: \"Which runs are waiting for my approval?\"\n- Approve: \"Approve the two flagged onboarding exceptions.\"\n- Describe a new agent: \"Build an agent that checks every supplier invoice against its contract.\"\n\nHow it works\nDescribe the asset and work you want to automate, and PitCrew MCP takes in a request and creates an agent and asset for you. Quotes, proposals, and forms are all created in your firm's branding. Send as many as you need, each with a personalized link and message. All PitCrew assets are AI-enabled and offer agentic support from the get-go. Every asset is password-protected and built to your firm's policies. BYOC and VPC deployments are available on request, so customer data can stay in your own cloud.\n\nReady-made agents cover sales, onboarding, service, billing, data and reporting work. PitCrew connects to common CRMs, document stores, spreadsheets and email, and the same run can be started from Slack, Outlook or Gmail as well as ChatGPT.\n\nSetup\nSign in with your PitCrew account and you are all set. Your data is never used to train models at PitCrew.",
"developerName": "PitCrew",
"category": "Business & Operations",
"websiteURL": "https://www.gopitcrew.com/",
"privacyPolicyURL": "https://www.gopitcrew.com/privacy",
"termsOfServiceURL": "https://www.gopitcrew.com/terms",
"supportURL": "https://www.gopitcrew.com/docs/mcp",
"defaultPrompt": [
"List my PitCrew agents",
"Build an agent that summarises Form ADV filings",
"Schedule my Compliance Calendar Builder agent to run every weekday at 9am Pacific"
],
"logo": "./assets/icon.svg",
"composerIcon": "./assets/icon.svg",
"brandColor": "#d9512b",
"capabilities": [
"Build AI agents by chatting",
"Run agents on documents",
"Create forms and proposals",
"Schedule and trigger agents",
"Check status and approve"
]
},
"onboardingSkill": "./skills/orchestrator/SKILL.md",
"publication": {
"release_notes": "Production release. 10 skills plus MCP server integration for building and running compliance agents.",
"countries": []
},
"review": {
"test_cases": {
"positive": [
{
"description": "User lists existing PitCrew agents.",
"prompt": "List my PitCrew agents",
"tools_triggered": "list_agents",
"expected_behavior": "Plugin fetches the user's PitCrew agents and displays them as a numbered list showing agent name and status (published or draft)."
},
{
"description": "User describes a compliance workflow they want to automate. The plugin uses local skills and the tools/ catalog to plan and build the agent spec, then asks for approval before publishing.",
"prompt": "Build an agent that summarises Form ADV filings given a CRD number",
"tools_triggered": "list_toolkits",
"expected_behavior": "Plugin produces a plan for a Form ADV Summariser agent with tasks such as retrieving Parts 1 and 2, extracting disclosures, identifying risks, and generating a report. Shows a link to review the full spec and asks the user to approve and provide a CRD number."
},
{
"description": "User approves the plan and provides the sample CRD. Plugin publishes the agent and runs it end-to-end with in-chat status polling.",
"prompt": "Approve. CRD number: 361",
"tools_triggered": "publish_agent, start_agent_run, get_run_status",
"expected_behavior": "Plugin publishes the Form ADV Summariser to the user's PitCrew workspace, starts a run with CRD 361 (Goldman Sachs), polls status in-chat until complete, and returns the final summary covering Form ADV Parts 1 and 2, disclosures, and risk findings."
},
{
"description": "User re-runs the same agent on a different firm to confirm the agent generalises across CRDs.",
"prompt": "Run the same agent for CRD 128888",
"tools_triggered": "start_agent_run, get_run_status",
"expected_behavior": "Plugin starts a new run of the Form ADV Summariser with CRD 128888 as input, polls status, and returns the summary for that firm."
},
{
"description": "User asks for a specific slice of the generated report, showing the output is queryable in chat.",
"prompt": "What are the top 3 risks from the CRD 361 report?",
"tools_triggered": "get_run_status",
"expected_behavior": "Plugin retrieves the completed CRD 361 run output and lists the top three risk findings with brief explanations for each."
}
],
"negative": [
{
"description": "User asks to delete all their agents. The MCP server does not expose a delete_agent tool, so the plugin should decline and direct the user to manage deletions in the PitCrew dashboard.",
"prompt": "Delete all my PitCrew agents"
},
{
"description": "User asks to run an agent that does not exist in their workspace. Plugin should look it up, fail to find it, suggest the closest matches, and ask for the correct agent name or ID rather than guessing.",
"prompt": "Can you run the email classifier agent"
},
{
"description": "User provides an invalid (non-numeric) CRD. Plugin should reject the input and ask for a valid numeric CRD before starting a run.",
"prompt": "Run the Form ADV Summariser for CRD banana"
}
]
},
"commerce": false,
"commerce_description": "The plugin works with any PitCrew account, free or paid. Authentication is via OAuth. No in-plugin purchases, credits, or paid upgrades gated by the plugin.",
"demo_recording_url": "https://www.loom.com/share/120c3994a0bf4f448ff16d62e540e75f"
}
}
}
}
Client extensions
Data this package carries for particular clients. The directory lists the clients named and never reads what is addressed to them.
- com.openai