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
3 commits sit in this repo's top risk tercile, which is 20% of the 15scored. 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 9.8 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.
Began feature-led, now leaning other.
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 | 500dc1c3Graph compiler now materializes typed action-rule, approval, evaluator, receipt, guard-join, and acceptance paths. Durable scheduler and transition monitor make invalid controller transitions fail closed. Added causal event storage, per-iteration node IDs, instance-bound receipts, semantic evidence graphs, replayable TLA+ counterexamples, signed execution certificates, proof-carrying workspace transitions, and signed paired harness experiments. Added focused runtime tests for joins, approval nod | Dinmay Kumar Brahma | 3w ago | +2.2K -64 | 90thElevated | more lines added than baseline |
| 2 | f3fded3aHarden durable production workflow services | Dinmay Kumar Brahma | 1mo ago | +2.9K -110 | 90thElevated | more lines added than baseline |
| 3 | 31311732Implement verified adaptive loop agent | Dinmay Kumar Brahma | 1mo ago | +3.7K -0 | 90thElevated | more lines added than baseline |
| 4 | 02b59b55External graph nodes now require signed, role-owned completions bound to graph, contract, run, node instance, artifacts, and expiry. Scheduler uses durable reserve → complete transitions; certificates replay that same lifecycle. Split snapshot request/materialization and executor dispatch/result paths; removed target-payload success fabrication. Compiled concrete evaluator/receipt nodes per protected evaluation plan and bound signed receipts to exact graph instances. Added dynamic policy/approva | Dinmay Kumar Brahma | 3w ago | +1.2K -329 | 63thTypical | more lines added than baseline |
| 5 | 34c10ae8mplemented the audit’s P0 fixes plus baseline deployment hardening. Idempotency now rejects key reuse across different intent digests. Approval and reconciliation are resumable workflow phases. Protected evaluator orchestration is wired into controller/runtime verification. Safety gates, global final guards, and probes cover multi-step acceptance. Memory promotion uses aggregate completion evidence and all run evidence refs. Profiles bind risk/probe metadata; production validates probe-policy di | Dinmay Kumar Brahma | 1mo ago | +1.8K -128 | 63thTypical | more lines added than baseline |
| 6 | e6922131Production configuration gate: [runtime.py](/data/ssd2/v_loop/src/vloop/runtime.py) rejects insecure executors, unsigned required checks, missing durable stores/probes/final verifier, missing signed approvals, and mismatched policy/executor keys. Receipt schema v2 binds canonical artifact manifest, primary artifact, contract, workspace snapshot, evaluator image/test-suite policy, key ID, and revocation epoch in [receipts.py](/data/ssd2/v_loop/src/vloop/receipts.py). Crash-safe SQLite idempotency | Dinmay Kumar Brahma | 1mo ago | +1.7K -132 | 63thTypical | more lines added than baseline |
| 7 | f5ca857dImplemented the review’s P0 security corrections plus key P1 hardening. | Dinmay Kumar Brahma | 1mo ago | +1.6K -133 | 63thTypical | more lines added than baseline |
| 8 | 53efe3ecadd | Dinmay Kumar Brahma | 1mo ago | +1.4K -38 | 63thTypical | more lines added than baseline |
| 9 | 6d9d936e ailed executions cannot advance criteria, reach final acceptance, or enter completion evidence. Prepared operations are deterministic, persisted before effects, and reconciled only via operation-bound signed receipts. Production requires the concrete Firecracker reconciler and prevents dependency overrides. Probe digests bind complete immutable manifests; evaluator plans are matched to receipt policies. Firecracker remote calls use a canonical execution-spec digest including workspace snapshot | Dinmay Kumar Brahma | 1mo ago | +877 -68 | 43thTypical | more lines added than baseline |
| 10 | a8233954Typed graph runtime now statically checks authority chains, guard dominance, recovery, cycles/timeouts, provenance flow, memory authority, and human escalation. Prepared effects, Firecracker specs/reconciliation, evaluator requests, and signed receipts are graph/node-bound. Added constrained read-only dynamic subgraphs, governed harness evolution with independent promotion/rollback, and model-controlled topology benchmarks. | Dinmay Kumar Brahma | 3w ago | +1.0K -10 | 37thTypical | more lines added than baseline |
| 11 | 5c6c1248mplemented the Rust-native trusted-kernel foundation without breaking the Python API. What moved to Rust when installed: SHA-256 over canonical bytes Ed25519 signing/verification for capabilities, approvals, receipts, artifacts, completions, and certificates Strict key/signature validation and native API version checks | Dinmay Kumar Brahma | 3w ago | +945 -49 | 30thBelow typical | more lines added than baseline |
| 12 | 2288daaaImplemented the Phase 1 graph-runtime foundation | Dinmay Kumar Brahma | 4w ago | +369 -20 | 23thBelow typical | more lines added than baseline |
| 13 | fe690439Reorganized the public API around clear trust-boundary folders: vloop.control — contracts, policy, controller, runtime, recovery vloop.evidence — attestations, snapshots, receipts, ledger, certificates vloop.execution — executors, Firecracker, supervisor integration vloop.intelligence — planner, context, memory, probes, repair, shadow model vloop.governance — harness evaluation, promotion, rollback, delegation | Dinmay Kumar Brahma | 3w ago | +344 -25 | 17thBelow typical | more lines added than baseline |
| 14 | 4229736bChange badge link to Ask DeepWiki | DINMAY KUMAR BRAHMA | 4h ago | +1 -1 | 10thBelow typical | fewer lines added than baseline |
| 15 | 76c70a34Update README.md | DINMAY KUMAR BRAHMA | 2w ago | +0 -0 | 3thBelow 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 15 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.
dino65-dev/v-loop has 15 commits in its history from 2 contributors, the first of them Jul 14, 2026. In the last 90 days 53 files were touched, 173 times in total, most often controller.py. Every commit is scored for change risk from its size, spread and the history of the files it touches.
