# Brainstorm: Local-first command analytics

## Metadata

- Date: 2026-06-17
- Seed: How can local command usage show which research workflows need better automation?
- Source labels allowed: user-idea, model-suggestion, citation-backed

## User Ideas

- [user-idea] Operators repeat the same manual CLI sequences when a framework
  lacks a good high-level skill.
- [user-idea] Failed commands may be more useful than successful commands for
  prioritizing documentation.

## Model Suggestions

- [model-suggestion] Group usage by project-relative path so workflows in
  `.aiwg/research/` can be separated from general maintenance commands.
- [model-suggestion] Compare high-frequency command clusters against existing
  skill trigger phrases.

## Citation-Backed Context

- None yet. This note should not be treated as a research claim.

## Open Questions

- Which commands signal research discovery versus research induction?
- What privacy constraints should apply before any heatmap feature exists?

## Links

- Concepts: [[local-first telemetry]], [[research workflow]]
- Entities: [[AIWG CLI]]
- Prior brainstorms: []

## Promotion Candidates

- Research question: Which local usage signals predict missing research skills?
- Acquisition target: examples of privacy-preserving local analytics in CLI tools