Python CLI
The Python CLI (toothfairy-cli) exposes 70 commands covering the core
platform surface. This page walks through the most common workflows; for the
complete flag-by-flag reference, see the
Python CLI command reference.
Install:
pip install toothfairy-cli
toothfairy configure --api-key YOUR_KEY --workspace-id YOUR_WORKSPACE
Send a message to an agent
toothfairy send "Hello" --agent-id AGENT_ID
# continue an existing chat
toothfairy send "Follow-up" --agent-id AGENT_ID --chat-id CHAT_ID
Stream a response
toothfairy send-stream "Explain RAG" --agent-id AGENT_ID
Predictions
toothfairy predict --model sorcerer --message "Summarise this" \
--max-tokens 256 --temperature 0.3
predict requires --modelUnlike the Node CLI, the Python predict command requires the --model flag
explicitly.
Agents CRUD
toothfairy create-agent --label "Support" --mode retriever \
--interpolation-string "You are a helpful assistant." --goals "Help customers"
toothfairy list-agents
toothfairy get-agent AGENT_ID
toothfairy update-agent AGENT_ID --label "Support v2"
toothfairy delete-agent AGENT_ID
Documents
toothfairy upload ./report.pdf
toothfairy get-doc DOC_ID
toothfairy delete-doc DOC_ID
Knowledge search
toothfairy search "How do I configure SSO?"
Scheduled jobs & triggers
toothfairy create-scheduled-job --agent-id AGENT_ID --frequency daily --prompt "Daily summary"
toothfairy list-scheduled-jobs
toothfairy pause-scheduled-job JOB_ID
toothfairy create-trigger --agent-id AGENT_ID --name "Webhook trigger" \
--event-source-type EXTERNAL
toothfairy list-triggers
Benchmarks
toothfairy create-benchmark --name "Retrieval quality"
toothfairy run-benchmark BENCHMARK_ID
toothfairy list-benchmark-runs
Parity gaps vs the Node CLI
The Python CLI does not implement voice, entities, folders, prompts, members, sites, hooks, or secrets commands. For those resources, use the Node CLI or the Python SDK.
Tips
- Add
--helpto any command for full flags, e.g.toothfairy create-agent --help. - The shorter alias
tfworks everywhere:tf list-agents.