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
11 commits sit in this repo's top risk tercile, which is 34% of the 32scored. 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 7.9 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 | Review priority | Top driver |
|---|---|---|---|---|---|---|
| 1 | 12e7d5e1WikiSkill for Hermes: harness + CLI + framework skills + demo bench + tests | wikiskill | 4d ago | +2.0K -0 | 98thElevated | more lines added than baseline |
| 2 | d8a55423Agent backends: Claude Code alongside Hermes (issue #13) (#14) | Ashutosh Singh | 4d ago | +1.0K -187 | 95thElevated | more lines added than baseline |
| 3 | fc067f11feat(backends): Codex + GitHub Copilot CLI backends (closes #15, #24) (#27) | Ashutosh Singh | 1d ago | +684 -3 | 92thElevated | more lines added than baseline |
| 4 | efb462c3Creative GitHub Pages redesign (custom animated landing) (#18) | Ashutosh Singh | 4d ago | +300 -7 | 84thElevated | more lines added than baseline |
| 5 | 249ea758compare: paired statistical run comparison (issue #3) (#6) | Ashutosh Singh | 4d ago | +282 -2 | 84thElevated | more lines added than baseline |
| 6 | 1beb1451Fix phantom-grading bug: fresh sandbox per rollout (force materialize wipes stale deliverables); warn on zero-tool-call sessions; --model/--provider patch isolated profile default instead of broken -m routing (strip moa 'default' fallback) | Ashutosh Singh | 4d ago | +248 -136 | 84thElevated | more lines added than baseline |
| 7 | dbe3178bPolish for public release: fame-grade README, docs/RUNS.md (documented inference), MIT license, CONTRIBUTING, CI workflow, pyproject metadata, lint cleanup, CI-readiness fix | Ashutosh Singh | 4d ago | +347 -86 | 84thElevated | more lines added than baseline |
| 8 | cdda5647feat: Hermes skills tap — 6 SKILL.md skills distilled from real evolution runs (wikiskill-evolve meta-skill + 5 operational patterns); README tap section (#26) | Ashutosh Singh | 2d ago | +268 -0 | 77thElevated | more lines added than baseline |
| 9 | 14a97034transfer: cross-workspace skill export/import (issue #1) (#7) | Ashutosh Singh | 4d ago | +162 -1 | 73thElevated | more lines added than baseline |
| 10 | 7a3b2663docs site: mkdocs-material GitHub Pages (issue: repo polish) (#8) | Ashutosh Singh | 4d ago | +175 -0 | 70thElevated | more lines added than baseline |
| 11 | 9c999954Add trap task family (subtle specs, debug tasks) + --max-turns/--run-budget knobs | wikiskill | 4d ago | +133 -8 | 67thElevated | more lines added than baseline |
| 12 | e75fd118fix(compare): stop forwarding max_turns to gating.run_task (TypeError on every real run) (#23) | lonylonyday-cyber | 2d ago | +70 -5 | 64thTypical | more lines added than baseline |
| 13 | 9f3f975fFix sandbox anchoring: absolute WORKING DIRECTORY in prompt (--in doesn't pin cwd); defensive grading (missing deliverable = 0); overwrite propagation for crash-resume | wikiskill | 4d ago | +47 -29 | 61thTypical | more lines added than baseline |
| 14 | 8f4ce01fRun 5 (gemini-lite) results: baseline 0.6667, harmful skill rejected (0.4444); wiki audit-trail commits (skeleton/maintainer/gate) | Ashutosh Singh | 4d ago | +67 -5 | 58thTypical | more lines added than baseline |
| 15 | 4e45f93aMulti-bug debug traps (non-compensating) + --no-early-stop demo knob + stay-in-sandbox instruction | wikiskill | 4d ago | +40 -22 | 55thTypical | more lines added than baseline |
| 16 | bebf7859docs: document the phantom-grading incident (Run 4) and the gemini-lite acceptance run (Run 5) | Ashutosh Singh | 4d ago | +31 -8 | 52thTypical | more lines added than baseline |
| 17 | 3dab4972release: auto-build sdist/wheel on tags; GitHub Release with artifacts; PyPI publish when PYPI_TOKEN configured (#9) | Ashutosh Singh | 4d ago | +31 -0 | 48thTypical | more lines added than baseline |
| 18 | 3301418edocs: Run 6 — compounding experiment (issue #5): honest negative result (3 iters, zero accepted skills, proposer no_action; accumulation needs a stronger model) | Ashutosh Singh | 4d ago | +30 -0 | 44thTypical | more lines added than baseline |
| 19 | c430651eEmbed full rejected-proposal content in skill-impact.md (paper fidelity); maintainer/proposer budgets 60 turns; selective trace reading | wikiskill | 4d ago | +23 -11 | 44thTypical | more lines added than baseline |
| 20 | 4cb78c00docs(tap): honest paper-alignment framing (maintainer patterns vs gate-accepted skills) + per-skill install commands | Ashutosh Singh | 2d ago | +20 -13 | 38thTypical | more lines added than baseline |
| 21 | b3a17ed5assets: social banner (#10) | Ashutosh Singh | 4d ago | +23 -0 | 38thTypical | more lines added than baseline |
| 22 | d4072d2efix: ship wikiskill.backends in the wheel (published 0.1.1 was broken — cli.py crashed on import); bump 0.1.2; release workflow smoke-tests the wheel before publishing; README: PyPI badge/install, roadmap done+planned; .env gitignored (#22) | Ashutosh Singh | 4d ago | +30 -11 | 33thBelow typical | more lines added than baseline |
| 23 | 76745fd6identity: rename package to wikiskill (multi-agent — hermes is the reference backend, claude ships, codex/opencode on roadmap); repo URLs updated | Ashutosh Singh | 4d ago | +13 -13 | 30thBelow typical | more scattered than baseline |
| 24 | d0ff7146release: PyPI trusted publishing (OIDC) — id-token: write, no API-token secret, publish step unconditioned on tags (#20) | Ashutosh Singh | 4d ago | +5 -6 | 27thBelow typical | fewer lines added than baseline |
| 25 | 3b30f8affix(release): GitHub parser-safe workflow — block-style tags, no braces in if-expression (#11) | Ashutosh Singh | 4d ago | +4 -3 | 22thBelow typical | fewer lines added than baseline |
| 26 | 5d116bb8README: correct gemma run to invalid/phantom-graded; document --model/--provider semantics | Ashutosh Singh | 4d ago | +4 -3 | 22thBelow typical | fewer lines added than baseline |
| 27 | aba9760bfix(release): secrets context not allowed in step if: — map to job env instead (actionlint-verified) (#12) | Ashutosh Singh | 4d ago | +4 -2 | 17thBelow typical | fewer lines added than baseline |
| 28 | 7aeb21cdBench: both debug tasks in val split (14 train / 8 val) | wikiskill | 4d ago | +3 -3 | 14thBelow typical | fewer lines added than baseline |
| 29 | 02fac2c8chore: bump to 0.1.4 (codex + copilot backends from #27) (#28) | Ashutosh Singh | 1d ago | +2 -2 | 8thBelow typical | fewer lines added than baseline |
| 30 | 70158c81chore: bump to 0.1.3 (compare max_turns fix from #23) (#25) | Ashutosh Singh | 2d ago | +2 -2 | 8thBelow typical | fewer lines added than baseline |
| 31 | a93b9c95feat/version 011 (#21) | Ashutosh Singh | 4d ago | +2 -2 | 8thBelow typical | fewer lines added than baseline |
| 32 | 4f10b3fdfix(tap): skills must live under skills/ for hermes skills tap discovery | Ashutosh Singh | 2d ago | +0 -0 | 2thBelow 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 32 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.
ashutoshsinghpr7/wikiskill has 32 commits in its history from 3 contributors, the first of them Aug 29, 2026. In the last 90 days 25 files were touched, 71 times in total, most often cli.py. Every commit is scored for change risk from its size, spread and the history of the files it touches.
