Every commit scored for change-risk against this repo's own history, so 'elevated' means elevated here rather than on some global curve.
Needs review
6 commits sit in this repo's top risk tercile, which is 33% of the 18scored. The cut is drawn against this codebase's own history rather than a global curve, so a quiet repo still fills its top band, and here it starts at 6.9 out of 10. What pushes a commit up is size and spread together: a large change confined to one area scores below a smaller one scattered across a dozen files.
Commit categories over time, read off the subject line. Fixes carry the accent because that is the series this chart exists to show.
How this repo's commit mix shifts over time.
Ranked by change-risk, highest first. Priority is a tercile of this repo's own distribution, so a quiet repo still fills its top band.
| # | Commit | Author | When | Lines | Risk | Top driver |
|---|---|---|---|---|---|---|
| 1 | 6acc352esync from context-xray @ b73ce48: LiteLLM sole price source | Ryan Li | 2mo ago | +42.5K -1.6K | 94%Elevated | more lines added than baseline |
| 2 | 71dbce58cost-xray — decompose your coding agent's context window and cost from the wire | Ryan Liclaudeassisted | 2mo ago | +10.0K -0 | 94%Elevated | more lines added than baseline |
| 3 | 60c068f2feat(install): macOS launchd supervisor (login-start + auto-restart) | Ryan Liclaudeassisted | 2mo ago | +337 -99 | 86%Elevated | more lines added than baseline |
| 4 | 137c0c24fix(tui): main-line context snapshot, prune, materialize cap, sticky port | Ruihan Liclaudeassisted | 1mo ago | +220 -12 | 81%Elevated | more lines added than baseline |
| 5 | e660c9b4fix(materializer): launch the warm consumer as a subprocess, not a multiprocessing spawn | Ryan Liclaudeassisted | 2mo ago | +74 -63 | 75%Elevated | more lines added than baseline |
| 6 | b15311cbinstall: harden run.sh against fresh-install + non-login environments | Ryan Liclaudeassisted | 2mo ago | +54 -2 | 69%Elevated | more lines added than baseline |
| 7 | 16660b42fix(adapters): codex session naming skips startup probe and injected wrappers | ryan | 2mo ago | +48 -29 | 64%Typical | more lines added than baseline |
| 8 | 53cba8c2fix(install): macOS launchd — nohup-start now when the agent can't load headless | Ryan Liclaudeassisted | 2mo ago | +32 -9 | 58%Typical | more lines added than baseline |
| 9 | 29d1e175sync from context-xray @ 6ce4b28 | Ryan Li | 2mo ago | +21 -1 | 53%Typical | more lines added than baseline |
| 10 | 91ea46c0fix(install): portable sed in the wrapper strip — macOS reinstall was aborting | Ryan Liclaudeassisted | 2mo ago | +14 -2 | 47%Typical | more lines added than baseline |
| 11 | d832da59docs: state Claude tokenization is calibrated estimation; announce opt-in exact mode | ryan | 2mo ago | +11 -4 | 42%Typical | more scattered than baseline |
| 12 | bb43cac6fix(install): macOS wrapper self-heal — delegate cold-path restart to run.sh start | Ryan Liclaudeassisted | 2mo ago | +9 -20 | 36%Typical | fewer lines added than baseline |
| 13 | 485fa189README: tighten the intro + drop the CX_OFF one-off note | Ryan Liclaudeassisted | 2mo ago | +3 -3 | 31%Below typical | fewer lines added than baseline |
| 14 | 19b23a6bdocs(readme): reframe intro around request-time context; drop tokenizer badge | ryan | 2mo ago | +3 -4 | 25%Below typical | fewer lines added than baseline |
| 15 | bfe83759sync from context-xray @ 58d46be | Ryan Li | 2mo ago | +2 -2 | 19%Below typical | fewer lines added than baseline |
| 16 | 3798d44esync from context-xray @ e653d89: install reinstall idempotency | Ryan Li | 2mo ago | +1 -1 | 11%Below typical | fewer lines added than baseline |
| 17 | 4e4e795ffix(run.sh): pick the port with the venv Python, not system python3 | Ryan Liclaudeassisted | 2mo ago | +1 -1 | 11%Below typical | fewer lines added than baseline |
| 18 | 236bd739docs: update detail screenshot | ryan | 2mo ago | +0 -0 | 3%Below typical | fewer lines added than baseline |
Two views of the same model: where the cuts fall, and what commit shape lands you above them.
Every scored commit, binned on the raw 0 to 10 score rather than the percentile. Percentile ranks are uniform by construction, so that axis has no shape to draw. The dashed lines are the tercile cuts behind each row's priority pill.
The 18 most recent commits, on their own recency sample rather than the feed above: that defaults to risk-sorted, so reusing it would plot only the top tercile and call it the spread. Big and scattered is what the model penalises. Click a dot to open it.
Repowise tracks change history across 46 files in 2233admin/cost-xray. In the last 90 days 46 files were touched, 110 times in total, most often run.sh. Every commit is scored for change risk from its size, spread and the history of the files it touches.
