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
81 commits sit in this repo's top risk tercile, which is 34% of the 237scored. 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 7.8 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 | 16506461v0.20.0: --pack (single-shot context engine) + benchmark on jcodemunch's corpus + pip packaging | Nehal | 3d ago | +3.4K -46 | 100%Elevated | more lines added than baseline |
| 2 | 6581c4c6v0.78.0: loop closure — --verify-edit oracle, --blindspot guard, --savings-report (local, no telemetry), memory growth bounds (200KB caps + lossless archive + --memory-prune), --eval runner plumbing, --install-hook warn-only; MCP 77→79 tools + loom://resources; 94 tests | Nehal | 23h ago | +1.3K -10 | 99%Elevated | more lines added than baseline |
| 3 | 4bc8cc98v0.79.0: Memory OS — typed memory.jsonl objects (type/id/importance/confidence/tier/affected_symbols), deterministic importance engine, --memory <symbol> graph-linked retrieval (entries + graph neighbors), --memory-add/--memory-stats, dual-write markdown+JSONL, MCP 79→82 tools (memory_add/remember/stats/prune); 101 tests | Nehal | 16h ago | +945 -32 | 99%Elevated | more lines added than baseline |
| 4 | dae2fa34memory-os: deterministic git-history extractor (bug/api/architecture heuristics, dedupe, --dry-run), memory_eval benchmark (10/10 hits, 84.4% fewer tokens synthetic / 99.2% fastapi), CI memory-eval step | Nehal | 16h ago | +837 -1 | 98%Elevated | more lines added than baseline |
| 5 | ea1e7608v0.77.0: README translations (zh/ja/es/hi) + benchmark-driven docs; version bump | Nehal | 1d ago | +914 -1 | 98%Elevated | more lines added than baseline |
| 6 | 0425b3aefix: MCP full dispatch parity (41 dead tools live), --resume restores decisions, --checkpoint --task writes, security (JSON-only index, --verify exit 1, path containment, auto-grammars opt-in), --lessons/--goal flags; tests 77->83 with MCP-registry + CLI dispatch coverage | Nehal | 1d ago | +479 -128 | 97%Elevated | more lines added than baseline |
| 7 | 3bf6c079bench: vs code-review-graph side-by-side + compaction-recovery benchmark (97.9% fewer tokens) | Nehal | 1d ago | +453 -5 | 97%Elevated | more lines added than baseline |
| 8 | 6df2a64ev0.7.0: close every accuracy/scalability weakness + honest limits | Nehal | 3d ago | +595 -87 | 97%Elevated | more lines added than baseline |
| 9 | f7ef2af1codemap: zero-dependency codebase outline generator for AI agents | Nehal | 3d ago | +538 -0 | 97%Elevated | more lines added than baseline |
| 10 | 1a52b15dbench+ci: eval_runner one-command suite (token/compaction/sealed/bench, --json, loss rows), oath job (stdlib-only check), bench-proof job (artifact), warn-only pre-commit hook, ignore .codeloom/ | Nehal | 23h ago | +601 -8 | 96%Elevated | more lines added than baseline |
| 11 | bc3401fcfeat: task-shaped tools + freshness envelope (repowise parity, plan WS-A) | Nehal | 1d ago | +411 -1 | 96%Elevated | more lines added than baseline |
| 12 | f0a5dd60Optional C-accelerated core (codeloom_core.c) + --engine c | Nehal | 2d ago | +427 -5 | 96%Elevated | more lines added than baseline |
| 13 | 10eca770v0.59.0: agent working-memory journal + layered compaction survival | Nehal | 2d ago | +426 -30 | 93%Elevated | more lines added than baseline |
| 14 | b992c1b3v0.37.0: all remaining 7 features — refactor engine, bug prediction, timeline, session dedupe, plugin SDK, LSP bridge, visual graph | Nehal | 3d ago | +399 -3 | 93%Elevated | more lines added than baseline |
| 15 | 640b807bv0.33.0: loom_context intent engine + repository memory + git intelligence | Nehal | 3d ago | +422 -4 | 93%Elevated | more lines added than baseline |
| 16 | 58dd6edaAdd rich documentation: CAPABILITIES, USER_GUIDE, CLI, SECURITY | Nehal | 3d ago | +347 -0 | 93%Elevated | more lines added than baseline |
| 17 | 3e20242aRename codemap -> codeloom (name was overused) | Nehal | 3d ago | +467 -467 | 93%Elevated | more lines added than baseline |
| 18 | 90fd142av0.14.0: token-shaving retrieval (byte offsets + token counts) + token benchmark | Nehal | 3d ago | +389 -13 | 93%Elevated | more lines added than baseline |
| 19 | c490c8d9v0.6.0: cross-file call graph + symbol index/search (deep AST analysis) | Nehal | 3d ago | +367 -6 | 93%Elevated | more lines added than baseline |
| 20 | e5fe94f6v0.5.0: task-aware intelligence (--task, --impact, --plan) + benchmarks | Nehal | 3d ago | +365 -5 | 93%Elevated | more lines added than baseline |
| 21 | 392d2bd2v0.4.0: --diff, multi-language call graph, --install-agents, --cost | Nehal | 3d ago | +327 -16 | 93%Elevated | more lines added than baseline |
| 22 | 0bd85798v0.3.0: function-level call graph + zero-dep MCP server + demo GIF | Nehal | 3d ago | +400 -11 | 93%Elevated | more lines added than baseline |
| 23 | 482650d4v0.63.0: full standalone Rust rewrite (codeloom_rs) | Nehal | 2d ago | +525 -3 | 90%Elevated | more lines added than baseline |
| 24 | d7cc8710v0.13.0: --explain, --similar, --deadcode (understanding & refactoring) | Nehal | 3d ago | +326 -14 | 90%Elevated | more lines added than baseline |
| 25 | ae426fd0v0.9.0: optional progressive-enhancement backends (overcome the limits) | Nehal | 3d ago | +423 -31 | 90%Elevated | more lines added than baseline |
| 26 | 35dbb37ev0.2.0: import dependency graph + --focus (real structural intelligence) | Nehal | 3d ago | +263 -10 | 90%Elevated | more lines added than baseline |
| 27 | 76df8bb9bench: regenerate RESULTS.md with real measured data (was --help dump), vs_crg real timing (no hardcoded 0.0), token_efficiency auto-clones repos, sync numbers to live runs | Nehal | 1d ago | +279 -98 | 88%Elevated | more lines added than baseline |
| 28 | 55c9d4e1chore: audit-fix batch from competitor-gap plan | Nehal | 1d ago | +540 -9 | 88%Elevated | more lines added than baseline |
| 29 | eed17e98v0.62.0: real Rust accelerator core (codeloom_core_rs) | Nehal | 2d ago | +387 -28 | 88%Elevated | more lines added than baseline |
| 30 | 71822d41Phases 3+5+6: CI action, load-once benchmark, launch material | Nehal | 3d ago | +294 -36 | 88%Elevated | more lines added than baseline |
| 31 | ca3e6577v0.10.0: snippet search, persistent cache, trust/credibility | Nehal | 3d ago | +326 -17 | 88%Elevated | more lines added than baseline |
| 32 | d4aeca03v0.68.0: kill the four slowness axes | Nehal | 1d ago | +288 -31 | 86%Elevated | more lines added than baseline |
| 33 | 38d0d066v0.36.0: architecture detection + dependency heatmap + explain-topic + auto-docs | Nehal | 3d ago | +270 -4 | 86%Elevated | more lines added than baseline |
| 34 | 3b4d3231docs: Memory OS v0.79 — typed memory objects, importance formula, graph-linked --memory retrieval, extractor, MCP 82 tools; translations + LAUNCH synced | Nehal | 16h ago | +262 -64 | 86%Elevated | more lines added than baseline |
| 35 | d7ac6336Add side-by-side token benchmark: codeloom vs jcodemunch (same repo, same symbols) | Nehal | 3d ago | +174 -0 | 86%Elevated | more lines added than baseline |
| 36 | 1d1a3405v0.76.0: binary-document extraction — docx/xlsx/pptx/epub/odt/rtf -> searchable sidecars | Nehal | 1d ago | +237 -3 | 84%Elevated | more lines added than baseline |
| 37 | 2ccc939efeat: PR review bot — risk verdict + health + reviewer brief, clean-room | Nehal | 1d ago | +184 -0 | 84%Elevated | more lines added than baseline |
| 38 | 8397a1a8v0.65.0: 16-language tree-sitter + neural-embedding infra + distro docs | Nehal | 2d ago | +250 -15 | 84%Elevated | more lines added than baseline |
| 39 | c5381684v0.53.0: framework routes + pub-sub channels + portable graph export | Nehal | 2d ago | +323 -13 | 84%Elevated | more lines added than baseline |
| 40 | 53a0da69The moat: --pack is now a code-embedded task brief + edit-relevance ranking | Nehal | 3d ago | +194 -52 | 84%Elevated | more lines added than baseline |
| 41 | 96236946v0.16.0: fix correctness gaps (nested gitignore, cache invalidation, workspace roots, trace safety, grammar installer) | Nehal | 3d ago | +230 -23 | 84%Elevated | more lines added than baseline |
| 42 | 1289acfev0.8.0: search upgrade — usages, snippets, multi-language index | Nehal | 3d ago | +201 -49 | 84%Elevated | more lines added than baseline |
| 43 | 55dfb0dcdocs: field matrix updated with code-review-graph/CCE/claude-context/codeseek (crg measured live); sealed live-agent harness | Nehal | 1d ago | +188 -38 | 81%Elevated | more lines added than baseline |
| 44 | f38710c3v0.72.0: structural pattern search — the feature neither competitor ships | Nehal | 1d ago | +207 -4 | 81%Elevated | more lines added than baseline |
| 45 | 03b4fc7av0.70.0: change-risk engine + live-freshness one-liner + watcher latency floor | Nehal | 1d ago | +224 -19 | 81%Elevated | more lines added than baseline |
| 46 | bb1c8c44v0.64.0: multi-repo graph + real tree-sitter engine + measured kernel benchmark | Nehal | 2d ago | +272 -8 | 81%Elevated | more lines added than baseline |
| 47 | fba1dbabv0.60.0: rich long-term memory + cognitive-load layer + session-aware search | Nehal | 2d ago | +270 -14 | 81%Elevated | more lines added than baseline |
| 48 | 80f44320docs: rewrite README — cleaner structure, all current information | Nehal | 1d ago | +140 -919 | 79%Elevated | more lines added than baseline |
| 49 | ac21c5b9v0.67.0: grammar parity fixtures + static-replay bench harness (plan WS-B2/C1) | Nehal | 1d ago | +382 -9 | 79%Elevated | more lines added than baseline |
| 50 | 2d31687av0.34.0: hybrid search + session memory | Nehal | 3d ago | +187 -8 | 79%Elevated | more 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 200 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.
sloemo01/codeloom has 241 commits in its history from 2 contributors, the first of them Aug 19, 2026. In the last 90 days 56 files were touched, 303 times in total, most often codeloom.py. Every commit is scored for change risk from its size, spread and the history of the files it touches.
