ai-tool-access-request-workflow
Sets up a complete intake-to-approval pipeline for AI tool access requests. Creates a Jira issue type with structured fields, configures automation rules for risk-based tier assignment and reviewer ro
SKILL.mdskills/ai-tool-access-request-workflow/SKILL.md
GitHubAI Tool Access Request Workflow
Standardized workflow for employees to request AI tool access with auto-routing to security review.
Integrations: Jira, Slack
When to Use
- The user wants to set up an approval process for AI tool requests
- An organization needs to standardize how employees get access to AI tools
- The user mentions "tool access request", "approval workflow", or "AI tool provisioning"
- Security or IT teams need a structured intake process
Steps
Step 1: Define Approval Tiers
Help the user define risk-based tiers:
| Tier | Risk Level | Examples | Approval Required |
|---|---|---|---|
| Tier 1: Pre-Approved | Low | GitHub Copilot, Cursor (with SSO) | Auto-approved, manager notification |
| Tier 2: Standard Review | Medium | ChatGPT Team, Claude, custom MCP servers | Manager + Security review (SLA: 3 days) |
| Tier 3: Extended Review | High | Tools handling PII/PHI, self-hosted LLMs, fine-tuned models | Manager + Security + Legal + CISO (SLA: 10 days) |
Step 2: Create Jira Request Template
Set up a Jira issue type "AI Tool Access Request" with these fields:
| Field | Type | Required |
|---|---|---|
| Tool Name | Short text | Yes |
| Tool URL/Vendor | URL | Yes |
| Business Justification | Long text | Yes |
| Data Types Involved | Multi-select (None, Internal, Confidential, PII, PHI) | Yes |
| Team/Department | Select | Yes |
| Number of Seats | Number | Yes |
| Estimated Monthly Cost | Number | No |
| Urgency | Select (Standard, Expedited) | No |
Step 3: Configure Jira Automation
Set up automation rules:
- Auto-assign tier — Based on "Data Types Involved" selection, auto-label the tier
- Route to reviewers — Tier 1: auto-approve and notify manager. Tier 2: assign to security team queue. Tier 3: create sub-tasks for each reviewer
- SLA tracking — Set due dates based on tier SLA
- Escalation — Auto-escalate if SLA is breached (notify reviewer's manager)
Step 4: Set Up Slack Notifications
Configure Slack integration:
- #ai-tool-requests channel: Post new requests and status updates
- DM to requester: Notify when status changes (approved, needs more info, denied)
- DM to reviewers: Alert when a new request needs their review
- Weekly digest: Summary of pending requests to the security team channel
Step 5: Create Runbook
Provide a runbook for the security review team:
- Vendor assessment checklist — SOC2 compliance, data retention policy, encryption, subprocessors
- Data flow diagram template — Where does data go when using this tool?
- Decision matrix — Standard criteria for approve/deny/conditional approval
- Conditional approval template — Approved with restrictions (e.g., no PII, VPN required)
- Denial template — Professional denial with alternative tool suggestions
Output
The workflow produces:
- A Jira project board with the request intake process
- Slack channel integration for visibility
- Automation rules for routing and SLA tracking
- Templates for security review decisions