CLI
Agent
Run and continue model-led Ploid research from the terminal.
Run a task
ploid ask "Research Retool and return three stage-matched lookalikes with evidence."Give the model the outcome and evidence standard. Use a larger budget only when the work genuinely needs deeper or broader research.
ploid ask "Research these seed accounts and return stage-matched lookalikes" \
--max-acu 1.6 \
--max-output-tokens 12000Continue a session
ploid ask "Research Retool and three stage-matched lookalikes" --session-id retool_01
ploid ask "Now rank them by buyer and stage fit" --session-id retool_01| Option | Purpose |
|---|---|
--session-id <id> | Continue a logical conversation |
--max-acu <0.2-6.4> | Set the API admission/billing limit |
--max-output-tokens <256-64000> | Set echoed compatibility metadata |
--memory <none|read|read_write> | Choose the long-term memory policy |
The CLI returns conversational Agent output. Use the Agent API structured-output contract when downstream code needs a caller-defined JSON Schema.
Errors and limits
Agent jobs use the same workspace credits, key budgets, and organization/API-key request buckets as POST /v1/agent. The CLI's default client timeout is 90 seconds; put --timeout <seconds> before ask for deeper work, or use --timeout 0 to disable the client deadline. API failures exit with status 2. See CLI errors & rate limits.