
Where did your Claude Code agent's time go? Reads local transcripts, shows tests, builds, deploys
Where did your Claude Code agent's time go? Reads local transcripts, shows tests, builds, deploys, polling, model time and waiting for you in an HTML report, and suggests fixes.
·
A Claude Code skill that shows where the wall-clock time and quota of your agent runs went (tests, builds, deploys, polling, model time, waiting for you) and ranks the fixes.
For anyone running long or autonomous Claude Code sessions who wants the next run to be faster and cheaper. Free, MIT licensed, runs locally.
npx skills add jindraregal/analyse-bottleneck # this project
npx skills add jindraregal/analyse-bottleneck --global # or every projectOr as a plugin:
/plugin marketplace add jindraregal/analyse-bottleneck # inside Claude Code
/plugin install analyse-bottleneck@jindraregal
claude plugin marketplace add jindraregal/analyse-bottleneck # or the same from a shell
claude plugin install analyse-bottleneck@jindraregalThen ask Claude Code: "use analyse-bottleneck on my last 24 hours". Needs Python 3.8+, standard library only.
It reads the transcripts Claude Code already writes to ~/.claude/projects, streaming (a few hundred MB take seconds). About 90% of the work is plain Python: every second of wall-clock time goes to exactly one cause, parallel tools and subagents are never double counted, and waits on subagents and background tasks are resolved to what they were doing. The model only reads a compact brief (<= 60 lines) and adds judgement, so it is cheap on a tight weekly quota.
view): a donut of agent time by cause plus a "Waiting for you" slice (your next message, permission prompts, approval gates), with ranked "Do this" fixes and the exact commands behind each slice.rules/*.json: Next.js, npm vs pnpm/bun, pip vs uv, jest vs vitest, tsc incremental, Docker cache, Playwright reruns, ffmpeg, whisper and more. Add your own with --extra-rules DIR.calls (every tool call and model turn), compare (two windows or sessions, to prove a fix helped), dora (delivery metrics from git and exported CI runs), watch (opt-in, non-blocking Stop hook; never installed for you).--sysstat, --jobs and scripts/sample_machine.py turn inferred machine signals into measurements.## Where the time went
| category | time | % of autonomous | % of span |
| human_wait:prompt | 15.0m | - | 50.6 |
| build | 8.6m | 58.7 | 29.0 |
| poll_wait | 4.0m | 27.4 | 13.5 |
1. **Full `next build` used as a quick correctness check** - ~6.0 min, confidence medium
- Arithmetic: 6 runs, 8.6m measured x 0.70 = 6.0m
- Before: `npm run build # after every small edit` After: `npx tsc --noEmit --incremental && npx next lint`S=~/.claude/skills/analyse-bottleneck/scripts
python3 $S/analyze.py view --project my-app --since 7d --out ~/rb --name my-app # .html, .brief.md, .view.json
python3 $S/analyze.py compare --project my-app --before "2026-10-01..2026-10-10" --after "2026-10-10.."Full options and the model's flow are in SKILL.md; changes per version in CHANGELOG.md.
Transcripts hold no CPU, memory or disk telemetry, so machine signals are inferred unless you pass measurements. Savings are estimates with stated assumptions you can edit in the rules. Everything runs locally: no network calls, no credentials read, nothing uploaded. Reports go to ~/.local/share/analyse-bottleneck/reports/, outside any git checkout.
Background: My AI coding agents got slower. Turns out the mess was mine.
MIT. Tests: python3 -m unittest discover -s tests (synthetic fixtures only).
Paste this prompt into your agent. It reads this page and does the setup for you.
Read https://aiagentslisting.com/mcp/analyse-bottleneck to learn what the "Analyse Bottleneck" MCP server does and how to install it. Add it to my coding agent's MCP configuration as documented on that page, then confirm the server connects and list the tools it exposes.Agents can also browse this directory over MCP at https://aiagentslisting.com/api/mcp
npx skills add jindraregal/analyse-bottleneck # this project
npx skills add jindraregal/analyse-bottleneck --global # or every projectThis server runs locally, so we can't read its tool list over the web yet.
We could not run this server: it ships no npx package for our sandbox and no public endpoint. The four checks that need a live server are left out, and the grade tops out at B until one exists.
Maintain this server? Claiming proves you own the listing and earns the last 4 points.
Claim this listingPaste this line near the top of your repository's README. The badge always renders the current score.
[](https://aiagentslisting.com/mcp/analyse-bottleneck)Nothing comparable is listed yet.
No tagged releases on record.
Connect to the AI Agents Listing MCP server at https://aiagentslisting.com/api/mcp using the HTTP transport and complete the sign-in when it asks. Then call the submit_review tool for the listing "analyse-bottleneck" with a rating from 1 to 5 and a short note on what worked and what did not.No reviews yet
Be the first to share your experience with this tool.
One email a week. New agents, MCP servers and skills, and what is actually getting traction.