task_charts__waterfall__running_total_extremum_value

Contract

1. Domain: charts
2. Scene id: waterfall
3. Source implementation: src/trace_tasks/tasks/charts/waterfall/running_total_extremum_value.py
4. Public query ids: maximum_running_total, minimum_running_total
5. Prompt query key: running_total_extremum_value

Implementation

1. Registered class: trace_tasks.tasks.charts.waterfall.running_total_extremum_value.ChartsWaterfallRunningTotalExtremumValueTask
2. Prompt bundle: src/trace_tasks/resources/prompts/charts/waterfall/charts_waterfall_v1.json
3. Generation is deterministic from instance_seed, explicit params, prompt bundle, renderer config, and code versions.
4. Answers and annotation are produced from the same metadata execution trace.

Annotation Contract

1. Answer schema: integer_value.
2. Annotation schema: bbox.
3. The annotation marks the full start or contribution bar where the requested cumulative extremum is reached.
4. The summary final bar is not a candidate because it duplicates the last contribution's running total.

Program Contract

Program: extremum(running_total(start_and_contribution_bars), direction); direction={maximum,minimum}; output=integer_value; annotation=bbox(extremum_bar); scene=waterfall; scope=running_total_extremum_value

Candidate set: the visible waterfall bars, contribution steps, and total bars inside the running_total_extremum_value objective scope. Operands: prompt-bound labels, categories, series names, thresholds, intervals, references, and encoded chart values, plus the active query id's comparator, direction, target role, or extremum focus when present. Operation: evaluate extremum over the candidate set using the filters, comparisons, aggregations, rankings, projections, or counterfactual edits named in the program expression; generation enforces a unique final answer. Output binding: answer is the integer_value value bound by integer_value. Annotation witnesses: bbox witnesses bound by bbox(extremum_bar). The annotation marks the full start or contribution bar where the requested cumulative extremum is reached. The summary final bar is not a candidate because it duplicates the last contribution's running total. Query ids: maximum_running_total, minimum_running_total.

Reasoning Operations

Families: ranking, aggregation

Query Details

Query id	Program contract	Answer schema	Annotation schema
maximum_running_total	max(running_total(start_and_contribution_bars)); output=integer_value; annotation=bbox(extremum_bar); scene=waterfall; scope=running_total_extremum_value	integer_value	bbox
minimum_running_total	min(running_total(start_and_contribution_bars)); output=integer_value; annotation=bbox(extremum_bar); scene=waterfall; scope=running_total_extremum_value	integer_value	bbox