Tools provided
Tools the agent exposes — your AI client calls them automatically when it needs them.
- compare_attribution_models — Report how far each channel's credit MOVES across ${CONVENTIONS.length} attribution conventions, because the spread is the uncertainty. None of them measures causation - they are rules for dividing a number that already happened, and touchpoint data cannot say which division is causal. A channel credited 31 percent under one rule and 9 under another was never measured at 31. Under ${ROBUST_SPREAD_POINTS} points with an unchanged rank it is ROBUST. Nothing is averaged: a mean would be a fifth convention hiding the only measured quantity.
- read_funnel — Say what a PERFECT fix at each funnel step could be worth in final conversions, and rank by that rather than by drop rate. A step losing 60 percent of 100 people loses 60; one losing 10 percent of 10,000 loses 1,000 - and the rate ranking puts the first on top. A fix is also bounded by what follows: rescuing people into a path that converts at 4 percent is worth 4 percent of them. No drop is called a leak, because whether a drop is a fault is a question about intent that counts cannot answer.
- read_retention_curve — Say how much of your retention curve every cohort actually reached, and where the evidence behind it thins. In any retention table the later periods are visible only to older cohorts: period 1 may rest on twenty cohorts and period 12 on one, printed in the same row at the same weight. Each period reports HOW MANY cohorts could observe it, and the last period every cohort reached is named as the honest end. Nothing is fitted, smoothed or projected.
- run_full — Run every capability you supplied the input for, in registry order, and return one report. Capabilities whose input is missing are listed as skipped with the fields they need — never run against a fixture and never reported as zero.
- list_capabilities — List what this agent can do: every tool it advertises, what each one is for, and the arguments each one reads. Useful for discovery before the first call.
- plan_inputs — Plan the inputs for a tool: returns the questions to ask, the JSON schema of the arguments, and a ready-to-edit example. Every argument it names is an argument the tool actually reads.
- discover_intent — Understand your goal and co-design the exact input through clarifying questions before anything runs.
- open_form — Use this the moment the user wants to run this agent on their own input. It opens an interactive form that collects the run arguments. Do NOT check credentials, do NOT ask clarifying questions in chat, and do NOT run any analysis before calling this — the form is the entry point for running the agent.
- run_form — Internal: invoked by the open_form form when the user submits. Reads exactly the same arguments as run_full and forwards all of them. Not for direct use.
