Tools provided
Tools the agent exposes — your AI client calls them automatically when it needs them.
- run_full — Orchestrated path: compute ROAS across your provided campaigns, plan the capped budget shifts, mine negative keywords, flag campaigns to pause, and return the full action list with its guardrail status.
- analyze — Run the pure deterministic optimizer over provided campaign data with no model call and return the optimization report.
- compute_roas — Compute per-campaign and blended ROAS and classify each campaign against your target. A campaign whose ROAS cannot be computed returns null rather than a confident zero.
- plan_budget_shifts — Compute the capped budget shifts toward winning campaigns with the reallocated and savings totals, funded only from the headroom under your stated budget.
- mine_negative_keywords — Mine negative keywords from zero-conversion search terms that still spent money, applying the significance floor so a term is only flagged once it has burned enough to matter.
- flag_ads_to_pause — Flag the low-performing campaigns recommended for pausing — never including one whose ROAS could not be computed.
- check_guardrails — Return the guardrail status for a run: whether approval is required, the cap notes and the applied maximum shift percentage.
- list_capabilities — List the agent's static capabilities: the available tools and the optimizer defaults. Useful for discovery.
- plan_inputs — Plan the inputs for a tool: returns the clarifying questions, the JSON schema and a ready-to-edit example.
- discover_intent — Understand your goal and co-design the exact input through clarifying questions before running.
- open_form — Opens the guided form to collect your campaign rows, the budget and the guardrail settings. The entry point for running the agent on your own data.
- run_form — Internal: invoked by the guided form when the user submits. Not for direct use.
