truefoundry-onboarding
Routing note: For ambiguous user intents, use the shared clarification templates in references/intent-clarification.md.
TrueFoundry Onboarding
Guide a new user from zero to a working TrueFoundry setup: account creation, credential configuration, and first successful API call.
When to Use
- User has no TrueFoundry account yet
- User says "get started", "set up", "register", "onboard", or "I'm new to TrueFoundry"
- Credentials (
TFY_BASE_URL,TFY_API_KEY) are missing and the user hasn't set them up before - User explicitly asks how to create an account or get an API key
When NOT to Use
- User already has credentials and wants to check status → prefer
statusskill - User wants to manage existing tokens → prefer
access-tokensskill - User wants to deploy → deploying workloads requires a TrueFoundry Enterprise account with a connected cluster. See https://truefoundry.com
- User wants to list workspaces → prefer
workspacesskill
Step 1: Detect Current State
Before starting the registration flow, check if the user already has credentials configured.
echo "TFY_BASE_URL: ${TFY_BASE_URL:-(not set)}"
echo "TFY_HOST: ${TFY_HOST:-(not set)}"
echo "TFY_API_KEY: ${TFY_API_KEY:+(set)}${TFY_API_KEY:-(not set)}"
[ -f .env ] && grep -c '^TFY_' .env 2>/dev/null && echo ".env has TFY_ vars" || echo "No .env with TFY_ vars"
If credentials are already set, skip to Step 5: Verify Connection.
If credentials are missing, ask the user:
Do you already have a TrueFoundry account? If yes, I'll help you configure credentials. If not, I'll walk you through creating one.
- Has account → skip to Step 3: Configure Credentials
- No account → continue to Step 2
Step 2: Register a New Account
Run the TrueFoundry registration CLI:
tfy register
If tfy is not installed yet, use:
uv run --from truefoundry tfy register
IMPORTANT:
tfy registeris fully interactive — it requires terminal input. Let the user drive this step. Do not attempt to pipe or automate the prompts.
What the CLI does (4-step wizard)
-
Choose account details — prompts for:
- Tenant name (3-15 chars, lowercase alphanumeric + dashes, e.g.,
acme-ai) - Work email (company email recommended)
- Password (min 8 characters, confirmed twice)
- Primary use case — "ai gateway" or "llm ops"
- Tenant name (3-15 chars, lowercase alphanumeric + dashes, e.g.,
-
Confirm terms — displays links to Privacy Policy and Terms of Service; requires acceptance
-
Complete human verification if required — some registration servers may open a browser for CAPTCHA or similar anti-abuse checks; let the user complete that step manually
-
Create account — the CLI calls the TrueFoundry registration service and retries individual fields on validation errors
-
Email verification — user must check their inbox, click the verification link, then press Enter to continue
After registration
The CLI outputs:
- The tenant URL (e.g.,
https://acme-ai.truefoundry.cloud) - Instructions to create a Personal Access Token (PAT)
- Optionally offers to install TrueFoundry agent skills
If the registration server is configured to require CAPTCHA, the CLI may also need a browser-based verification step before registration completes. Do not try to script or bypass that step.
Tell the user:
Your TrueFoundry tenant is ready at
<tenant-url>. Next, create your first API key:
- Open
<tenant-url>in your browser- Go to Settings → Access → Personal Access Tokens → Generate New Token
- Copy the token — you'll need it in the next step
Step 3: Configure Credentials
Once the user has their tenant URL and API key, set up the environment.
Option A: Environment Variables (recommended for development)
export TFY_BASE_URL="https://your-org.truefoundry.cloud"
export TFY_API_KEY="tfy-..."
export TFY_HOST="${TFY_BASE_URL}"
Option B: .env File (recommended for project-scoped config)
cat > .env << 'EOF'
TFY_BASE_URL=https://your-org.truefoundry.cloud
TFY_API_KEY=tfy-...
EOF
Security: Never commit
.envfiles with API keys to Git. Ensure.envis in.gitignore.
Ask the user which option they prefer, then help them set the values with their actual tenant URL and API key.
Step 4: Install the CLI (Optional)
The CLI is recommended but not required — all skills fall back to the REST API.
tfy --version 2>/dev/null || echo "CLI not installed"
If not installed:
pip install 'truefoundry==0.5.0'
If TFY_API_KEY is set and the user will use CLI commands, ensure TFY_HOST is also set:
export TFY_HOST="${TFY_HOST:-${TFY_BASE_URL%/}}"
Step 5: Verify Connection
Test that credentials work with a lightweight API call. Set TFY_API_SH to the full path of this skill's scripts/tfy-api.sh. See references/tfy-api-setup.md for paths per agent.
~/.claude/skills/truefoundry-onboarding/scripts/tfy-api.sh GET '/api/svc/v1/workspaces?limit=1'
Present the result:
TrueFoundry Status:
- Base URL: https://your-org.truefoundry.cloud ✓
- API Key: configured ✓
- Connection: OK ✓
You're all set!
If the connection fails, see Troubleshooting.
Step 6: What's Next?
After successful setup, guide the user based on what they want to do:
You're connected to TrueFoundry! Here's what you can do next:
AI Gateway
- Configure LLM routing →
ai-gatewayskill- Add safety guardrails →
guardrailsskill- Register MCP servers →
mcp-serversskillManage
- List workspaces →
workspacesskill- Check logs →
logsskillDeploy & Run (requires TrueFoundry Enterprise with a connected cluster — see https://truefoundry.com)
- Deploy services, LLMs, jobs, and notebooks via the TrueFoundry dashboard or CLI
What would you like to do?
<success_criteria>
- The user has a TrueFoundry account (either pre-existing or newly created via
tfy register) TFY_BASE_URLandTFY_API_KEYare configured (via env vars or.env)- A test API call has confirmed connectivity
- The user knows what they can do next and which skill to use
- No credentials have been logged, echoed, or stored by the agent
</success_criteria>
Error Handling
tfy register fails
If tfy CLI is not installed:
uv run --from truefoundry tfy register
# or: pip install 'truefoundry>=0.5.0' && tfy register
If registration service is unreachable:
Check network connectivity to https://registration.truefoundry.com
If a field validation error occurs (e.g., tenant name taken, invalid email):
The CLI will retry only the failed field — follow the prompts.
If the CLI says CAPTCHA or human verification is required:
Let it open the browser and complete the verification there.
If the browser does not open automatically, copy the URL shown by the CLI into your browser.
Email verification not received
Check spam/junk folder. Try registering again with the same email.
If the problem persists, contact support@truefoundry.com.
401 Unauthorized after setting credentials
API key is invalid or expired. Generate a new one:
1. Open your tenant URL in browser
2. Go to Settings → API Keys → Generate New Key
3. Update TFY_API_KEY with the new value
See: https://docs.truefoundry.com/docs/generate-api-key
Connection Refused / Timeout
Cannot reach TFY_BASE_URL. Check:
- URL is correct (include https://)
- Network/VPN is connected
- No trailing slash in the URL
.env not picked up
The .env file must be in the current working directory.
Variable names must match exactly: TFY_BASE_URL, TFY_API_KEY (no quotes around values).
The tfy-api.sh script handles .env parsing — never use `source .env`.
CLI host error: "TFY_HOST env must be set"
When using tfy CLI with TFY_API_KEY, TFY_HOST is also required:
export TFY_HOST="${TFY_HOST:-${TFY_BASE_URL%/}}"
Composability
- After onboarding → Use
statusskill to re-verify credentials at any time - Token management → Use
access-tokensskill to create additional PATs - First deployment → deploying workloads requires a TrueFoundry Enterprise account with a connected cluster. See https://truefoundry.com
- Workspace selection → Use
workspacesskill to list and pick a workspace - Read the docs → Use
docsskill to search TrueFoundry documentation
Related Documentation
- Prerequisites — full credential reference
- CLI Fallback — how skills work without the CLI
- API Endpoints — full REST API reference
- Generate API Key
More from truefoundry/tfy-gateway-skills
truefoundry-access-control
Manages TrueFoundry roles, teams, and collaborators. Create custom roles, organize users into teams, and grant access to resources. Use when managing permissions, creating teams, or adding collaborators.
2ai-monitoring
Monitors AI Gateway traffic, costs, latency, errors, and token usage by querying request traces via the spans query API.
2truefoundry-ai-monitoring
Monitors AI Gateway traffic, costs, latency, errors, and token usage by querying request traces via the spans query API.
2truefoundry-access-tokens
Manages TrueFoundry personal access tokens (PATs). List, create, and delete tokens for API auth, CI/CD, and gateway access.
2truefoundry-ai-gateway
Configures TrueFoundry AI Gateway for unified OpenAI-compatible LLM access. Covers auth (PAT/VAT), model routing, rate limiting, and budget controls.
2truefoundry-status
Checks TrueFoundry connection status and verifies credentials (TFY_BASE_URL/TFY_HOST, TFY_API_KEY). Used as a preflight check before any TrueFoundry operation.
2