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
18 commits sit in this repo's top risk tercile, which is 35% of the 52scored. 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 9.0 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 | ab934794v1: async kernel, admission checker, sessions, HTTP API, tools, policy, ops | Shashank Shekhar Singhclaudeassisted | 1w ago | +25.9K -294 | 99%Elevated | more lines added than baseline |
| 2 | 7add137aClose the governed loop, and write a cookbook that runs | Shashank Shekhar Singhclaudeassisted | 1w ago | +14.6K -272 | 95%Elevated | more lines added than baseline |
| 3 | 76ee2685Harden the sandbox, make budgets real, give memory a disk, and build the AgentNode | Shashank Shekhar Singhclaudeassisted | 1w ago | +5.3K -189 | 95%Elevated | more lines added than baseline |
| 4 | 29a0dac8Implement GraphARC: a graph engineering toolkit on LangGraph | Shashank Shekhar Singhclaudeassisted | 1w ago | +5.8K -1 | 95%Elevated | more lines added than baseline |
| 5 | 841393b1Add a live trace view to serve, and a file-backed plan approval flow (#39) | Shashank Shekhar Singhclaudeassisted | 2d ago | +4.7K -118 | 90%Elevated | more lines added than baseline |
| 6 | 99d37f61Split the CLI surface, add configuration, and close the defects an audit found | Shashank Shekhar Singhclaudeassisted | 1w ago | +5.3K -2.2K | 90%Elevated | more lines added than baseline |
| 7 | 5c9a62a7Fix four metering/tracing defects, and colour the CLI without changing piped output (#11) | Shashank Shekhar Singhclaudeassisted | 6d ago | +1.5K -136 | 87%Elevated | more lines added than baseline |
| 8 | f767b6fdComplete governed planning CLI and documentation | Shashank Shekhar Singh | 1w ago | +2.0K -159 | 87%Elevated | more lines added than baseline |
| 9 | b524f30cAdd OpenRouter backend and a model registry | Shashank Shekhar Singhclaudeassisted | 1w ago | +1.2K -35 | 84%Elevated | more lines added than baseline |
| 10 | 723a177cRun the CLI from Slack over Socket Mode, behind an admission gate (#23) | Shashank Shekhar Singhclaudeassisted | 6d ago | +902 -3 | 82%Elevated | more lines added than baseline |
| 11 | 16094373Add an honest assessment and restate the vision as an agent runtime | Shashank Shekhar Singhclaudeassisted | 1w ago | +571 -0 | 80%Elevated | more lines added than baseline |
| 12 | 68c85e50Harden the gate against what an adversarial verifier actually broke | Shashank Shekhar Singhclaudeassisted | 1w ago | +678 -158 | 78%Elevated | more lines added than baseline |
| 13 | 39a8be7dFix five defects found by adversarial review of the harness, gateway, memory, and capstone | Shashank Shekhar Singhclaudeassisted | 1w ago | +437 -44 | 76%Elevated | more lines added than baseline |
| 14 | 37c6b41eExecute the serving cookbook's console transcripts, and mend their rot (#35) | Shashank Shekhar Singhclaudeassisted | 4d ago | +391 -17 | 72%Elevated | more lines added than baseline |
| 15 | 0da93a66Add a delegated executor: agent --executor claude-cli (#31) | Shashank Shekhar Singhclaudeassisted | 5d ago | +432 -13 | 72%Elevated | more lines added than baseline |
| 16 | 4270d0e7Add ARCHITECTURE.md: the golden-stage target architecture | Shashank Shekhar Singhclaudeassisted | 1w ago | +320 -0 | 72%Elevated | more lines added than baseline |
| 17 | c9dfb981An agent node on the Claude CLI now delegates instead of refusing (#58) | Shashank Shekhar Singhclaudeassisted | 1d ago | +414 -6 | 67%Elevated | more lines added than baseline |
| 18 | 53f7de07Ship a planning registry that reads, and let Slack name it (#28) | Shashank Shekhar Singhclaudeassisted | 6d ago | +374 -8 | 67%Elevated | more lines added than baseline |
| 19 | 96099420Bump to 0.1.1: fix __version__ drift and update PyPI install instructions | Shashank Shekhar Singhclaudeassisted | 6d ago | +577 -41 | 63%Typical | more lines added than baseline |
| 20 | 608fcfe4Pass the V0 gate: a real model fixes a bug and proves it with tests | Shashank Shekhar Singhclaudeassisted | 1w ago | +303 -10 | 63%Typical | more lines added than baseline |
| 21 | 731b018fAdd ROADMAP.md: the live build list from kernel to agent runtime | Shashank Shekhar Singhclaudeassisted | 1w ago | +233 -0 | 63%Typical | more lines added than baseline |
| 22 | 0ccc1a9bFix two false-reject bugs that only a live model run could expose | Shashank Shekhar Singhclaudeassisted | 1w ago | +238 -29 | 59%Typical | more lines added than baseline |
| 23 | 65f7fa0eAdd brand assets and put the logo at the top of the README (#54) | Shashank Shekhar Singhclaudeassisted | 2d ago | +250 -0 | 56%Typical | more lines added than baseline |
| 24 | feef03d7Initial commit | Shashank Shekhar Singh | 1w ago | +241 -0 | 56%Typical | more lines added than baseline |
| 25 | da8825fbA bare backend name was read as a model name, and the CLI's failures said nothing (#56) | Shashank Shekhar Singhclaudeassisted | 1d ago | +226 -3 | 53%Typical | more lines added than baseline |
| 26 | 9ea15bdbLet the operator open agent to Slack, behind a double opt-in (#27) | Shashank Shekhar Singhclaudeassisted | 6d ago | +182 -24 | 51%Typical | more lines added than baseline |
| 27 | 729c6c51Close a real sandbox escape: dir_fd= bypasses the workspace write grant (#53) | Shashank Shekhar Singhclaudeassisted | 2d ago | +142 -0 | 48%Typical | more lines added than baseline |
| 28 | a7fed812Add a propose kind: the model authors from what the registry read (#30) | Shashank Shekhar Singhclaudeassisted | 6d ago | +153 -32 | 48%Typical | more lines added than baseline |
| 29 | 6f9463e7Fix/backend name as model spec (#57) | Shashank Shekhar Singhclaudeassisted | 1d ago | +144 -162 | 45%Typical | more lines added than baseline |
| 30 | 932c409cClose a real sandbox escape on Python 3.12/3.13 that a brittle test was hiding | Shashank Shekhar Singhclaudeassisted | 1w ago | +108 -3 | 42%Typical | more lines added than baseline |
| 31 | 78dd36f1Refuse a Subgraph subclass outright; the first fix repelled the attack, not the mechanism | Shashank Shekhar Singhclaudeassisted | 1w ago | +120 -47 | 42%Typical | more lines added than baseline |
| 32 | a68d9575The README's first code block was LangGraph's API, not GraphARC's (#55) | Shashank Shekhar Singhclaudeassisted | 2d ago | +103 -17 | 39%Typical | more lines added than baseline |
| 33 | 46745fceAdd a command-by-command Slack walkthrough to the docs (#25) | Shashank Shekhar Singhclaudeassisted | 6d ago | +92 -0 | 38%Typical | more lines added than baseline |
| 34 | 2e5e5214Refuse a second session on another session's checkpoint thread (#34) | Shashank Shekhar Singhclaudeassisted | 4d ago | +86 -4 | 34%Typical | more lines added than baseline |
| 35 | b44099d1feat(cli): wire all Budget dimensions into grapharc run (#14) | Matthew | 6d ago | +90 -6 | 34%Typical | more lines added than baseline |
| 36 | 246c2544Rename ArcGraph/ArcState to GraphARC/GraphARCState for naming consistency | Shashank Shekhar Singhclaudeassisted | 1w ago | +86 -86 | 34%Typical | more lines added than baseline |
| 37 | 6f36c516Report a malformed trace line as the exit-2 document, not a traceback (#37) | Shashank Shekhar Singhclaudeassisted | 4d ago | +78 -8 | 30%Below typical | more lines added than baseline |
| 38 | 241b2724README refactoring: improve discoverability and structure | Shashank Shekhar Singhclaudeassisted | 6d ago | +67 -16 | 28%Below typical | more lines added than baseline |
| 39 | 9e7c68acAssert what the deadline guard guarantees, not where CPython delivers it | Shashank Shekhar Singhclaudeassisted | 6d ago | +57 -18 | 26%Below typical | more lines added than baseline |
| 40 | 99a5aca6Refuse a sync overrun at guard exit even when the timer never fired (#38) | Shashank Shekhar Singhclaudeassisted | 4d ago | +59 -6 | 22%Below typical | more lines added than baseline |
| 41 | 266d6a55Name the variable when a GRAPHARC_* env value cannot be coerced (#33) | Shashank Shekhar Singhclaudeassisted | 4d ago | +63 -1 | 22%Below typical | more lines added than baseline |
| 42 | a4235d79Re-derive every quoted claim now that four of the five seams are closed | Shashank Shekhar Singhclaudeassisted | 1w ago | +74 -67 | 22%Below typical | more lines added than baseline |
| 43 | 88bf1040Give a successful viz reply a link that renders the diagram (#24) | Shashank Shekhar Singhclaudeassisted | 6d ago | +61 -1 | 18%Below typical | more lines added than baseline |
| 44 | 633129f7Rank object spans ahead of array spans in JSON extraction (#32) | Shashank Shekhar Singhclaudeassisted | 4d ago | +34 -7 | 16%Below typical | more lines added than baseline |
| 45 | 3dbcce3dStop two timing-sensitive tests failing on a loaded CI runner | Shashank Shekhar Singhclaudeassisted | 6d ago | +18 -2 | 14%Below typical | more lines added than baseline |
| 46 | 857f6209Correct the quoted test count to 1328, and name the fifth gap in ARCHITECTURE §7 | Shashank Shekhar Singhclaudeassisted | 1w ago | +16 -8 | 13%Below typical | more lines added than baseline |
| 47 | 263c6c4dExport observe cost helpers (#1) (#12) | Vamsi Krishna | 6d ago | +12 -0 | 11%Below typical | more lines added than baseline |
| 48 | aea6ee72Release 0.1.2: bump the version everywhere the tree states it (#43) | Shashank Shekhar Singhclaudeassisted | 2d ago | +10 -10 | 9%Below typical | more scattered than baseline |
| 49 | c3d80014Embed the decomposition demo in the README, excluded from the sdist (#44) | Shashank Shekhar Singhclaudeassisted | 2d ago | +8 -0 | 7%Below typical | fewer lines added than baseline |
| 50 | 7a0284edRemove the memory extra: it installed neo4j with nothing behind it (#3) (#36) | Shashank Shekhar Singhclaudeassisted | 4d ago | +6 -36 | 5%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 52 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.
codegraphcontext/grapharc has 52 commits in its history from 3 contributors, the first of them Jul 27, 2026. In the last 90 days 170 files were touched, 389 times in total, most often main.py. Every commit is scored for change risk from its size, spread and the history of the files it touches.
