Display Name
retro-daily
Category
Hooks
Sub-Category
General
Primary Link
https://gyanesh-m.github.io/retro-daily/
Author Name
gyanesh-m
Author Link
https://github.com/gyanesh-m
License
MIT
Other License
No response
Description
A SessionStart hook that prints a daily retro at the top of every Claude Code session, competency grade (0–100, A–F), 14-day efficiency sparklines, year-long contributions heatmap. Spawns a detached, sandboxed claude -p background worker that researches your weakest metrics on docs.anthropic.com and GitHub and surfaces findings inline in the next session's dashboard.
Validate Claims
No response
Specific Task(s)
Try this single prompt in a fresh Claude Code session (with retro-daily installed):
Read the Focus Areas block from the retro-daily dashboard above.
For the metric flagged as my worst, read 3–5 recent transcripts under ~/.claude/projects/
and tell me concretely why I'm bad at it. End with a settings.json diff that addresses the root cause.
The dashboard turns "how do I actually get better at using Claude Code?" — a vague intent — into a transcript-grounded, single-prompt task with a working artifact at the end. That's the value: retro-daily doesn't just measure, it makes "improve myself" a concrete prompt Claude Code can act on.
Specific Prompt(s)
No prompt needed to trigger the dashboard, it renders automatically at the top of every Claude Code session as a SessionStart hook. Open a new claude session and the retro is already there.
The compounding value is when you use the dashboard data as input to your next prompt. Three to try:
Read my retro-daily dashboard at the top of this session. Pick the weakest metric from Focus Areas and walk me through 2–3 concrete settings.json or workflow changes that would improve it.Look at the SCOUT FINDINGS block. Pick the repo most relevant to my workflow and write me a one-paragraph evaluation of whether it's worth installing.My Auto-approve rate is below 80%. Show me the 5 tool prefixes I'm approving most often (check transcripts), then draft the settings.json entry to allow them automatically.
Additional Comments
I built this because I couldn't tell whether I was actually getting better at using Claude Code or just spending more on it — I had no way to find my weak points, let alone act on them. It's been the most useful thing in my own workflow since I shipped it; I look at the dashboard at the start of every session now.
Recommendation Checklist
Display Name
retro-daily
Category
Hooks
Sub-Category
General
Primary Link
https://gyanesh-m.github.io/retro-daily/
Author Name
gyanesh-m
Author Link
https://github.com/gyanesh-m
License
MIT
Other License
No response
Description
A SessionStart hook that prints a daily retro at the top of every Claude Code session, competency grade (0–100, A–F), 14-day efficiency sparklines, year-long contributions heatmap. Spawns a detached, sandboxed claude -p background worker that researches your weakest metrics on docs.anthropic.com and GitHub and surfaces findings inline in the next session's dashboard.
Validate Claims
No response
Specific Task(s)
Try this single prompt in a fresh Claude Code session (with retro-daily installed):
The dashboard turns "how do I actually get better at using Claude Code?" — a vague intent — into a transcript-grounded, single-prompt task with a working artifact at the end. That's the value: retro-daily doesn't just measure, it makes "improve myself" a concrete prompt Claude Code can act on.
Specific Prompt(s)
No prompt needed to trigger the dashboard, it renders automatically at the top of every Claude Code session as a
SessionStarthook. Open a newclaudesession and the retro is already there.The compounding value is when you use the dashboard data as input to your next prompt. Three to try:
Read my retro-daily dashboard at the top of this session. Pick the weakest metric from Focus Areas and walk me through 2–3 concrete settings.json or workflow changes that would improve it.Look at the SCOUT FINDINGS block. Pick the repo most relevant to my workflow and write me a one-paragraph evaluation of whether it's worth installing.My Auto-approve rate is below 80%. Show me the 5 tool prefixes I'm approving most often (check transcripts), then draft the settings.json entry to allow them automatically.Additional Comments
I built this because I couldn't tell whether I was actually getting better at using Claude Code or just spending more on it — I had no way to find my weak points, let alone act on them. It's been the most useful thing in my own workflow since I shipped it; I look at the dashboard at the start of every session now.
Recommendation Checklist