Failed to load insights data.
Check the server logs or try again.
No insights data available.
Data is generated nightly at 10 PM IST.
Overall Activity
PR Activity Over Time
Automation Adoption
Effort Distribution
Epics by Project
By Project
| Project | Created | Merged | Merge Rate | Auto Created | % Auto Created | Auto Merged | % Auto Merged | % Clean Merge | Time→PR (med) | Time→Merge (med) |
|---|
By Developer
| Developer | Created | Merged | Merge Rate | Auto Created | % Auto Created | Auto Merged | % Auto Merged | % Clean Merge |
|---|
Merge Rate Trend
Cost by Project
Automation by Agent Provider
| Provider | PRs Created | PRs Merged | Merge Rate | Clean Merge % |
|---|
Quality by Repo (Clean vs Dirty Merges)
Automation by Repo
| Repo | Generated | Merged | Merge Rate | Clean Merge % | Avg Dev Fixes |
|---|
Top Skills
| Skill | Invocations | Jobs | Merge Rate | Clean Merge % | vs Baseline |
|---|
Avg Tokens per PR
Top Token Consumers
Daily Token Usage by Type
Cache Hit Rate (Daily)
Tokens by Project
Token Breakdown by Project
Daily Breakdown
Sessions
Token Detail
Per-Phase Breakdown
Skills Invoked
| Skill | Calls | Output Tokens | Context Growth |
|---|
Tool Usage (Output Tokens)
Top Context Contributors (Tools, Skills, Prompts)
Turn Progression (Cumulative Tokens) — hover a point for tool / skill / context delta
Tool Usage Detail
| Tool | Calls | Output Tokens | % of Output | Context Growth |
|---|
Background Agents
| # | Tokens | In / CC / CR / Out | Cost | Turns | Top Tools |
|---|
No token analysis data available yet.
Data is collected nightly from Claude CLI session files.
Plan Stage Funnel
Ticket-grain. Same definitions as skill effectiveness.
Trends
Same definitions as the funnel above, bucketed by the ticket's first job. Rates come from summed counts per bucket — never an average of finer-grained percentages. Hover any point for its denominator; thin buckets are plotted, not hidden.
Plan Approval
Plan-Skip Share (minor scope)
Validation — Self-Reported
Merge Rate by Plan Arm
Planned vs Plan-Skipped
PR-producing tickets only. Not a controlled comparison — see the note below.
| Arm | Tickets | Merge | Clean of merged |
|---|
Plan Revision by Provider
Of plans entered. v1 = approved without revision.
| Provider | Plans entered | Revised | Revision rate | v1-approval | Skipped w/ PR |
|---|
Skill Effectiveness Roll-Up
per-skill detail on the Skills pageRanked by plan v1-approval delta vs the skill's own repo baseline — the most causally-clean signal. Same aggregator as the Skills page; no separate definition.
| Skill | Repo | Plan n | v1 Δ | PR n | Merge Δ | Clean Δ |
|---|
Validation — Agent Self-Report
These are claims the agent made about its own run, not independent verification. Treat as a self-report signal only.
How these are computed
Read with care
No plan-stage data for this window.
Requires DATABASE_URL and tickets that reached the plan stage.