Add date-range filtering to /insights
Feature Request
Add date-range or timeframe filtering to the /insights command so users can analyze specific windows of their usage history.
Why This Matters
The current /insights command analyzes all available sessions (~30 days) as a single block. But usage patterns, friction points, and workflows evolve over time. Being able to compare insights across different timeframes would surface habit changes — for example:
- "Am I hitting fewer hallucination issues this week vs. last month?"
- "Did adding that custom skill actually reduce my correction cycles?"
- "How has my workflow shifted since I started using MCP servers?"
Without timeframe filtering, there's no way to track whether workflow improvements are actually working or whether friction patterns are trending up or down.
Suggested API
# Filter by number of days
/insights --days 15
# Filter by date range
/insights --since 2026-02-13
/insights --from 2026-02-01 --to 2026-02-15
# Compare two periods
/insights --compare 15d # compares last 15 days vs. prior 15 daysBonus: Period Comparison
The most powerful version would support a --compare flag that runs insights on two adjacent time windows and highlights what changed — new friction patterns, resolved issues, shifts in project focus, etc. This would make /insights a true feedback loop for improving how users work with Claude Code.
7 Comments
+1 to this feature — it would make
/insightsmuch more useful for evaluating workflow changes over time.One workflow example:
I often experiment with Claude Code setups (adding MCP tools, changing prompts, updating skills). To see if those changes help, I want to compare productivity before and after the change.
Right now
/insightsaggregates about 30 days, which mixes multiple workflow iterations together.Being able to run something like:
/insights --since 2026-03-01
/insights --since 2026-03-08
/insights --compare 7d
would make it much easier to evaluate whether those changes actually improved efficiency.
One design question: should
--days Nrefer to calendar days or active-session days? Clarifying that behavior might help avoid confusion for users with intermittent usage.+1 on this. My
/insightsreport shows 1,825 total sessions but only 76 were analyzed — covering just ~16 days (March 17 – April 1). That's less than 5% of my history.For power users who've been on Claude Code for months, lifetime analysis would be significantly more valuable than a rolling 2-week window. The patterns, friction trends, and workflow evolution across months tell a much richer story than a snapshot.
Ideal behavior:
/insightswith no args → analyze all available session data (lifetime)/insights --since 2wor/insights --range 2026-03-01..2026-03-31→ scoped analysisThe data already exists locally in
~/.claude/usage-data/— it's just not being used fully.Just ran this command for the first time and it only picked my sessions from today, even though I have sessions going back at least 30 days. Would definitely like a date range filter. +1
+1 — would use this for quarter-over-quarter comparisons to tell whether CLAUDE.md changes and new
skills are actually reducing friction. Right now /insights flattens ~30+ days of sessions into a
single verdict, so incremental workflow changes are invisible. The --compare 15d variant in the
this would really be helpful especially valuable for A/B-ing custom skill rollouts and check if leads to desired result at scale across team members
Closing for now — inactive for too long. Please open a new issue if this is still relevant.
Wtf
This issue has been automatically locked since it was closed and has not had any activity for 7 days. If you're experiencing a similar issue, please file a new issue and reference this one if it's relevant.