Data & Analytics

Internal dashboard worker for the browser-based Brick Office view over .harness runtime state. Hire/call from CEO, CTO, CDO, CQO, or OPS; never as an Owner slash command.

Focused Signals scout for finding observability gaps in PostHog itself — significant event volumes the team isn't tracking, custom events with no insight or dashboard coverage, insights pointing at events that have stopped firing, dashboards missing related context, critical events with no alerts. Watches the event-stream-vs-saved- inventory delta as the team's product evolves and emits findings recommending new insights, dashboard additions, or alerts when gaps clear the confidence bar. Self...

Guides experiment state transitions: launching, pausing, resuming, ending, shipping variants, archiving, resetting, duplicating, and copying to another project. Covers preconditions, implications for variant assignment and analysis, and the decision framework for when to use each action.\nTRIGGER when: user asks to launch, pause, resume, end, ship, archive, reset, duplicate, or copy an experiment to another project.\nDO NOT TRIGGER when: user is creating an experiment (use creating-experiment...

Configures the analytics side of a PostHog experiment — exposure criteria (default `$feature_flag_called` vs custom exposure events), primary and secondary metrics, the supported metric types (count, sum, ratio with `math` and `math_property`, retention with `retention_window_start` and `start_handling`), multivariate user handling ("Exclude" vs "First seen variant"), and how to read results once the experiment is live. Use when the user adds or edits a primary or secondary metric (e.g. "add ...

Run the per-round tension check that makes /grill's steelman push-back discipline real. Reads the epic cut on the live model (the `epic.md` grounding docs, written in place per ADR 026 — there is no draft tree) and everything the slice declared — its functionalities' grounding docs `functionality.md` (the hub), all SEVEN lens grounding docs, and the profile bars — and produces a structured tension report, one entry per real contradiction, each citing the specific declared item it defends (sou...

Almost every downstream bug traces back to an index that was assumed regular and was not. Make the grid explicit before anything else.

Prepare temporal data for modeling without letting information from the future reach a model that will only ever see the past.

Customer name changes → Just update ```sql UPDATE dim_customer SET name = 'John Smith' WHERE customer_id = 123; ``` Pros: Simple | Cons: Lose history

An analytics implementation degrades along a predictable path: it starts clean, each team adds events in their own style, and within a year nobody can answer a question without first asking three people what an event means. The taxonomy is what prevents that, and it only works if it is enforced in code review.

Run this before anything else. It is the cheapest experiment in machine learning and it partitions the space of causes in one shot.

Explain what a trained model is doing — globally and for individual predictions — without mistaking the explanation for a statement about the world.

The search algorithm matters far less than the budget and the space you give it, but the choice still costs or saves hours.

"Feature importance" is ambiguous. Name the question before picking the tool.

Compress a wide feature space into fewer coordinates while keeping the structure that matters, and know what each method destroys in exchange.

A validation *boundary* is any point where data crosses from a system you do not control into one you do: an HTTP response, an S3 drop, a Kafka topic, a partner SFTP file, a form submission. Put one gate at each boundary that parses raw input into a typed domain object. Everything downstream of the gate may assume validity.

Catch bad data at the point it enters your system, so downstream tables, dashboards, and models fail loudly instead of quietly producing wrong numbers.

**Monthly Partitions:** ```sql CREATE TABLE events ( event_id BIGINT, user_id INT, event_time TIMESTAMP ) PARTITION BY RANGE (YEAR_MONTH(event_time)) ( PARTITION p202601 VALUES LESS THAN ('202602'), PARTITION p202602 VALUES LESS THAN ('202603'), PARTITION p202603 VALUES LESS THAN ('202604'), PARTITION p_future VALUES LESS THAN MAXVALUE ); ```

Almost every downstream bug traces back to an index that was assumed regular and

Run this before anything else.

The search algorithm matters far less than the budget and the space you give it,

\"Feature importance\" is ambiguous.

A validation *boundary* is any point where data crosses from a system you do not

**Monthly Partitions:**

This skill enables troubleshooting in the domain of time-series-analysis (data-science). It represents expert-level expertise and is designed for production use in research, industry, and educational contexts.