Analyze a user's cctrace usage and work patterns from read-only CLI JSON. Use for personal usage summaries, period comparisons, high-cost sessions, and practical workflow recommendations; never use it to retrieve transcripts or credentials.
Installs into .claude/skills of the current project.
Sponsors
SSkills DirectorySkills Directory
Know which skills are safe — weekly.
SSkills DirectorySkills Directory
Is that skill safe to install?
SSkills Directory ProSkills Directory
Get any skill into Claude in one click.
Are you the author of Usage Insights?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/wedatalab-ai-lab-usage-insights)
---
name: usage-insights
description: Analyze a user's cctrace usage and work patterns from read-only CLI JSON. Use for personal usage summaries, period comparisons, high-cost sessions, and practical workflow recommendations; never use it to retrieve transcripts or credentials.
---
# Usage Insights
Use the same workflow as [the shared agent skill](skills/usage-insights/SKILL.md). It is bundled here so this directory can be installed as a Codex skill.
Run only these read-only commands:
```sh
cctrace usage --since 7d --json
cctrace usage --since 14d --until 7d --json
cctrace ls --since 7d --limit 20 --json
cctrace tools --since 7d --json
cctrace plugins --since 7d --json
cctrace skills --since 7d --json
cctrace rules --json
cctrace organization-insights --since 7d --json
cctrace insights cost --since 7d --json
cctrace insights context --since 7d --json
```
For "why did my cost go up" and "am I wasting context", follow the scenario procedures in the shared skill and report every entry in `caveats`.
Report observations, interpretations, and recommended actions separately. Never print, request, or expose API tokens. Never use `cctrace sessions`; it can contain transcript records.
