f0abb2f · 23,618 commits · 8,367 files
Issued 2026-10-22
Reads strong: 4 of 5 rated dimensions are strong, 1 is adequate and none is significant.
105 checks ran across 6 assessable dimensions. Observability & Operations is adequate. 13 findings, each with evidence. One dimension cannot be assessed without an acquirer.
This repository reads strong. Show it in your README: the badge links to a page where anyone can check the reading. If you embed it: link it to the report, remove it if the report is delisted, and do not present it as a certification or an endorsement.
[](https://borehole.dev/verify/0S2YRN069R3T?via=badge)- 04.62 Releases are cut from another branch than the one surveyed
- states a policy on AI-generated contributions AGENTS.md:3
- 1 finding from checks Borehole's calibration found wrong more often than right is no longer published.
| Dimension | Rating | ||
|---|---|---|---|
| 01 | Architecture & Codebase — including AI provenance | strong | |
| 02 | Engineering Process & SDLC Maturity | strong | |
| 03 | Organization & Key-Person Risk | strong | |
| 04 | Product & Engineering Maturity | not assessable | |
| 05 | Security & Reliability Posture | strong | |
| 06 | Observability & Operations | adequate | |
| 07 | Fit with the Acquirer | not assessable |
- 84% of non-merge commits carry any trailer at all
- no widespread author/committer date divergence
- the declared share is neither decisive nor negligible, so the true figure is more sensitive to undeclared work
Architecture & Codebase — including AI provenance
01 · strong · ratings ↑rust/src/server/src/routes/tests.rs is 8,943 lines, 43× this codebase's median file
adequateThe median source file here is 206 lines. 50 files exceed 2,060 lines, the largest being 8,943. Size is not a defect on its own — some problems genuinely live in one place — but these are where merge conflicts, review fatigue and single-owner knowledge concentrate, and they are the first thing that slows an inheriting engineer down.
A note that does not affect this rating: measured against its own codebase's median, it is below the 85th percentile of 382 surveyed projects' largest files, where this check begins to count.
| line-count | rust/src/server/src/routes/tests.rs:1 — 8,943 lines (43× median) |
| line-count | vllm/v1/worker/gpu_model_runner.py:1 — 7,193 lines (34× median) |
| line-count | vllm/_aiter_ops.py:1 — 5,206 lines (25× median) |
| line-count | vllm/_custom_ops.py:1 — 4,865 lines (23× median) |
| line-count | vllm/v1/attention/ops/rocm_aiter_mla_sparse.py:1 — 4,757 lines (23× median) |
Respond to this finding
Your response is published beside the finding, not instead of it. A finding you can argue with is the point — and a wrong one is calibration data we want.
An owner's response is removed only when it is unlawful, impersonates someone, publishes personal data or a credential, or is not about the finding. Disagreeing with the report is never a reason. To report a response, or to reach us about one, write to takedown@borehole.dev.
Engineering Process & SDLC Maturity
02 · strong · ratings ↑Tests are retried until they pass
adequate34 tests are individually marked to retry. A retry turns a flaky test green without saying why it failed, so a green build means the tests passed eventually. Ask how many retries a typical run needs, and which tests need them.
A note that does not affect this rating: notes like it are common in well-run projects.
| tests/basic_correctness/models/test_basic_correctness.py | tests/basic_correctness/models/test_basic_correctness.py:83 — test retry |
| tests/entrypoints/multimodal/openai/chat_completion/test_video.py | tests/entrypoints/multimodal/openai/chat_completion/test_video.py:438 — test retry |
Respond to this finding
Your response is published beside the finding, not instead of it. A finding you can argue with is the point — and a wrong one is calibration data we want.
An owner's response is removed only when it is unlawful, impersonates someone, publishes personal data or a credential, or is not about the finding. Disagreeing with the report is never a reason. To report a response, or to reach us about one, write to takedown@borehole.dev.
Organization & Key-Person Risk
03 · strong · ratings ↑No finding. 10 checks ran and none fired; the rejection log below says why each did not.
Product & Engineering Maturity
04 · not assessable · ratings ↑Not rated: only 9 of 19 checks could reach a conclusion here; the rest do not apply to this repository or could not read it. The findings below are what it could still read.
Releases are cut from another branch than the one surveyed
adequateThe newest release, v0.32.0, is not on the branch this survey read; the newest release on it is v0.11.1. The shipped product may differ from what this report describes. Survey the release branch too.
A note that does not affect this rating: notes like it are common in well-run projects.
| 5df63a720 | tag v0.32.0 |
Respond to this finding
Your response is published beside the finding, not instead of it. A finding you can argue with is the point — and a wrong one is calibration data we want.
An owner's response is removed only when it is unlawful, impersonates someone, publishes personal data or a credential, or is not about the finding. Disagreeing with the report is never a reason. To report a response, or to reach us about one, write to takedown@borehole.dev.
Security & Reliability Posture
05 · strong · ratings ↑10 dependencies float, and no lockfile is committed
adequateDependency versions are declared as ranges and no lockfile is committed, so a build today does not necessarily produce what a build last month produced. That matters twice over in a transaction: the thing being bought cannot be reproduced exactly, and a compromised upstream release arrives without anyone choosing to take it.
A note that does not affect this rating: in Borehole's calibration, this check's notes were usually true but overstated.
| unpinned | benchmarks/kernels/requirements.txt:1 — any version |
| unpinned | benchmarks/multi_turn/requirements.txt:1 — >=1.24 |
| unpinned | benchmarks/multi_turn/requirements.txt:2 — >=2.0.0 |
| unpinned | benchmarks/multi_turn/requirements.txt:3 — >=3.10 |
| unpinned | benchmarks/multi_turn/requirements.txt:4 — >=4.46 |
Respond to this finding
Your response is published beside the finding, not instead of it. A finding you can argue with is the point — and a wrong one is calibration data we want.
An owner's response is removed only when it is unlawful, impersonates someone, publishes personal data or a credential, or is not about the finding. Disagreeing with the report is never a reason. To report a response, or to reach us about one, write to takedown@borehole.dev.
The build runs third-party code it does not pin
adequate1 third-party CI action is referenced by a tag or branch rather than a commit, so they run whatever that tag points to on the day; 2 steps pipe a download straight into a shell. Tags have been moved to malicious commits in real supply-chain attacks, and a piped script is whatever the server returns that minute. Pinning to a commit is the fix, and it is small.
A note that does not affect this rating: notes like it are common in well-run projects.
| .github/workflows/recipes.yml | .github/workflows/recipes.yml:25 — action pinned by tag |
| docker/Dockerfile | docker/Dockerfile:169 — download piped into a shell |
Respond to this finding
Your response is published beside the finding, not instead of it. A finding you can argue with is the point — and a wrong one is calibration data we want.
An owner's response is removed only when it is unlawful, impersonates someone, publishes personal data or a credential, or is not about the finding. Disagreeing with the report is never a reason. To report a response, or to reach us about one, write to takedown@borehole.dev.
10 outbound calls can wait forever
adequate10 HTTP calls use a client with no timeout (Python `requests` without `timeout=`, or Go's default client). When a dependency slows down, each one holds a worker until the connection gives up, and enough of them take the service down with the dependency.
A note that does not affect this rating: in Borehole's calibration, this check's notes were usually true but overstated.
| vllm/v1/core/sched/diffusion_scheduler.py | vllm/v1/core/sched/diffusion_scheduler.py:42 — outbound call without a timeout |
| vllm/v1/core/sched/scheduler.py | vllm/v1/core/sched/scheduler.py:2075 — outbound call without a timeout |
| vllm/v1/core/sched/scheduler.py | vllm/v1/core/sched/scheduler.py:2097 — outbound call without a timeout |
Respond to this finding
Your response is published beside the finding, not instead of it. A finding you can argue with is the point — and a wrong one is calibration data we want.
An owner's response is removed only when it is unlawful, impersonates someone, publishes personal data or a credential, or is not about the finding. Disagreeing with the report is never a reason. To report a response, or to reach us about one, write to takedown@borehole.dev.
2 shell commands built from strings
adequate2 places run a command through a shell (`shell=True`, `os.system`, `exec` with a template) rather than passing arguments. Where any part comes from outside, it is command injection. Ask where each one's input comes from.
A note that does not affect this rating: in Borehole's calibration, this check's notes were usually true but overstated.
| vllm/utils/cpu_resource_utils.py | vllm/utils/cpu_resource_utils.py:289 — command run through a shell |
| vllm/utils/cpu_resource_utils.py | vllm/utils/cpu_resource_utils.py:293 — command run through a shell |
Respond to this finding
Your response is published beside the finding, not instead of it. A finding you can argue with is the point — and a wrong one is calibration data we want.
An owner's response is removed only when it is unlawful, impersonates someone, publishes personal data or a credential, or is not about the finding. Disagreeing with the report is never a reason. To report a response, or to reach us about one, write to takedown@borehole.dev.
19 deserialisers that can run code
adequate19 places read data with a deserialiser that executes what it reads (pickle). That is safe on data the system wrote itself and remote code execution on anything else. Ask where each one's input comes from.
A note that does not affect this rating: in Borehole's calibration, this check's notes were usually true but overstated.
| vllm/compilation/caching.py | vllm/compilation/caching.py:130 — code-executing deserialiser |
| vllm/compilation/caching.py | vllm/compilation/caching.py:311 — code-executing deserialiser |
| vllm/distributed/device_communicators/all_reduce_utils.py | vllm/distributed/device_communicators/all_reduce_utils.py:417 — code-executing deserialiser |
Respond to this finding
Your response is published beside the finding, not instead of it. A finding you can argue with is the point — and a wrong one is calibration data we want.
An owner's response is removed only when it is unlawful, impersonates someone, publishes personal data or a credential, or is not about the finding. Disagreeing with the report is never a reason. To report a response, or to reach us about one, write to takedown@borehole.dev.
19 locked dependencies have published vulnerabilities
adequate19 of 631 locked dependencies match 22 published advisories in OSV. A lockfile says which version is installed, not whether the flawed code is reachable from this product, so each needs a look. Most are fixed by an upgrade, and the advisory names the version. This lockfile pins development tools alongside what ships and does not say which is which, so some of these may never reach a user.
A note that does not affect this rating: in Borehole's calibration, this check's notes were usually true but overstated.
| rust/Cargo.lock | rust/Cargo.lock — rustls 0.23.37: GHSA-2mjx-qc3c-rqvc (moderate), fixed in 0.23.45 |
| rust/Cargo.lock | rust/Cargo.lock — anyhow 1.0.102: RUSTSEC-2026-0190 (unrated), fixed in 1.0.103 |
| rust/Cargo.lock | rust/Cargo.lock — backoff 0.4.0: RUSTSEC-2025-0012 (unrated) |
Respond to this finding
Your response is published beside the finding, not instead of it. A finding you can argue with is the point — and a wrong one is calibration data we want.
An owner's response is removed only when it is unlawful, impersonates someone, publishes personal data or a credential, or is not about the finding. Disagreeing with the report is never a reason. To report a response, or to reach us about one, write to takedown@borehole.dev.
146 advisories against this project in the last 24 months, 56 high or critical
adequateThe newest was published on 2026-10-21. A project that publishes advisories is telling its users about its fixes, and the count and pace say how much security work it carries. Whether 52 of them are fixed in the code you ship could not be told from this history: its version is only known to be at least 0.11.1. Answer lives in: the release you run, set against each fixed version. GHSA-q97c-p368-75c4 (moderate), GHSA-6757-hggm-hp2w (high), GHSA-vfp2-c8pq-v6h6 (high) and 50 more are fixed in the surveyed commit; which release carries them cannot be told from this history, which does not place the release users run. Answer lives in: the version you run. The record is incomplete: NVD: 8 records name vllm with nothing tying them to this repository (CVE-2024-11041, CVE-2024-8939, CVE-2024-9053, …).
A note that does not affect this rating: it describes the project's advisory record, and no advisory in it is known to affect the surveyed commit.
| CVE-2026-114049 | 2026-10-21: CVE-2026-114049 (moderate): vLLM before 0.31.0 contains a cross-request mask leak in InputBatch.condense() that allows remote attackers to constrain other clients' outputs by leaving stale |
| CVE-2026-114051 | 2026-10-21: CVE-2026-114051 (high): vLLM before 0.31.0 skips the min_tokens <= max_tokens check for /inference/v1/generate requests with a null max_tokens, so the EngineCore input thread later die |
| CVE-2026-114052 | 2026-10-21: CVE-2026-114052 (high): vLLM after 0.24.0 and before 0.31.0 contains a memory exhaustion vulnerability in MOSS-Audio model serving that caches a processor for every distinct request-co |
| CVE-2026-114053 | 2026-10-21: CVE-2026-114053 (critical): vLLM from 0.7.3 before 0.31.0 forwards the request-supplied mm_processor_kwargs.code_revision to Transformers AutoProcessor.from_pretrained, letting clients sel |
| CVE-2026-114054 | 2026-10-21: CVE-2026-114054 (high): vLLM before 0.31.0 contains a cache aliasing vulnerability that allows authenticated attackers to collide shared cache keys by supplying caller-controlled value |
Respond to this finding
Your response is published beside the finding, not instead of it. A finding you can argue with is the point — and a wrong one is calibration data we want.
An owner's response is removed only when it is unlawful, impersonates someone, publishes personal data or a credential, or is not about the finding. Disagreeing with the report is never a reason. To report a response, or to reach us about one, write to takedown@borehole.dev.
Observability & Operations
06 · adequate · ratings ↑No structured logging library is in use
adequateNothing references a structured logging library. Logs that are written to be read by a person are not searchable by a machine, which means incident response here is proportional to how well someone remembers the codebase rather than to what the tooling can find.
| logging-scan | searched 6,620 files for 10 structured logging libraries; 0 matched |
Respond to this finding
Your response is published beside the finding, not instead of it. A finding you can argue with is the point — and a wrong one is calibration data we want.
An owner's response is removed only when it is unlawful, impersonates someone, publishes personal data or a credential, or is not about the finding. Disagreeing with the report is never a reason. To report a response, or to reach us about one, write to takedown@borehole.dev.
Sends usage data to stats.vllm.ai by default
adequatevllm/envs.py sends to stats.vllm.ai, and rust/src/server/src/routes/inference/generate/convert.rs leaves reporting on as shipped. Every installation reports to the vendor unless its operator turns this off. For a buyer running it inside their own product, that is an outbound data flow to a third party: ask what is sent, how often, and whether the buyer's customers were told.
| vllm/envs.py | vllm/envs.py:37 — reports to stats.vllm.ai |
| rust/src/server/src/routes/inference/generate/convert.rs | rust/src/server/src/routes/inference/generate/convert.rs:317 — the reporting setting is on by default |
Respond to this finding
Your response is published beside the finding, not instead of it. A finding you can argue with is the point — and a wrong one is calibration data we want.
An owner's response is removed only when it is unlawful, impersonates someone, publishes personal data or a credential, or is not about the finding. Disagreeing with the report is never a reason. To report a response, or to reach us about one, write to takedown@borehole.dev.
No error reporter is wired in
adequateNothing in the service initialises an error reporter (Sentry, Rollbar, Bugsnag, Honeybadger or an APM agent). An exception in production reaches a person only if someone is reading the logs, which usually means a customer reports it first.
A note that does not affect this rating: notes like it are common in well-run projects.
| error reporting | no error reporter initialised |
Respond to this finding
Your response is published beside the finding, not instead of it. A finding you can argue with is the point — and a wrong one is calibration data we want.
An owner's response is removed only when it is unlawful, impersonates someone, publishes personal data or a credential, or is not about the finding. Disagreeing with the report is never a reason. To report a response, or to reach us about one, write to takedown@borehole.dev.
Fit with the Acquirer
07 · not assessable · ratings ↑Not rated: structurally not assessable from a repository. What a person adds here.
Check it on the running system
8 checks against the running system follow from the findings above: what a buyer's engineer will try, how to run it first, and what to have ready. A paid survey of this repository includes them.
A paid survey of a public repository is also reviewed: a language model reads each finding against the code and withdraws only findings it judges the code contradicts, each listed with its reason. The review is a language model's reading of the findings against the code, and it can be wrong. It never adds a finding, never raises a band, and never removes a finding from a proven check or an advisory that affects the surveyed version.
What a person adds
The survey reads what the repository can show. The code cannot settle these questions, and on them this report states nothing. Each says where its answer lives: your own engineer, the company, or an assessor you choose can settle it. Questions marked measured open with something this repository shows.
Architecture & Codebase — including AI provenance
01 · full coverage- measured No architecture or design document is in the tree, across 4,407 product files. Where does the design live, and who holds it?
- Is the architecture right for the next three years of the product?
- Were the home-grown components reasonable choices when they were made, and are they now?
Engineering Process & SDLC Maturity
02 · high coverage- Is review real or a rubber stamp? Comment depth and time to approve need the pull request history.
- Is delivery planned and measured, or heroic?
- How are incidents run, reviewed and learned from?
Organization & Key-Person Risk
03 · partial coverage- measured 3,506 people have committed here. Which of them are you acquiring, and which are you replacing?
- Who mentors whom, and would that survive a departure?
- Can the team grow, and who stands behind the top three contributors?
- Is the seniority mix sound, and will pay survive your salary bands?
- Is the pace sustainable? Commit times cannot say: git keeps no reliable local time.
- Does the team want to stay after close?
- How are decisions made and disagreements resolved?
Product & Engineering Maturity
04 · minimal coverageNot rated: only 9 of 19 checks could reach a conclusion here; the rest do not apply to this repository or could not read it.
- measured 12,939 commits landed in the last year, against 7,569 the year before (+71%). What drove it: new people, a deadline, or code produced faster than before?
- measured Experiment tracking is configured (tensorboard). Can the model that ships be tied to the run, data and code that produced it?
- measured No model card is in the tree. What is the model's intended use, what was it trained on, and where does it fail?
- Can this team deliver the stated roadmap on the stated timeline?
- Does leadership know about the debt, and is time budgeted to repay it?
- Is the engineering shape consistent with the commercial story?
Security & Reliability Posture
05 · partial coverage- measured No threat model is in the tree of what looks like a running service. Does one exist elsewhere, and who owns it?
- What has actually broken, how often, and how badly?
- What did external testing find, and was it fixed?
- What is the SOC 2, ISO 27001 or GDPR status, and where is the evidence?
Observability & Operations
06 · partial coverage- measured Nothing in the tree backs up or restores data. How is data backed up, and has a restore ever been rehearsed?
- measured No alert rule is defined in the repository. Where are alerts defined, who receives them, and when did one last fire?
- Are the dashboards opened, the alerts acknowledged and the pages answered?
- Are the service-level objectives actually met?
- Who carries the pager, and how often does it go off?
- What does it cost to run, and how much headroom is there?
Fit with the Acquirer
07 · none coverageNot rated: structurally not assessable from a repository.
- This is primarily a Python codebase. Does your engineering organisation already run Python in production, or would this be the first?
- The bulk of this system sits in vllm/, csrc/, rust/. Which of your existing systems would each of those have to absorb or replace?
- The repository carries deployment configuration (docker/Dockerfile and 11 more). Does that match your deployment and on-call model, or does it need rehosting first?
- What is your tolerance for the findings above remaining unaddressed for the first two quarters after close, given everything else that will compete for this team's time?
- Which of these systems duplicate yours, and which get retired?
- What will integration cost, and in what order?
- Where does the team land in your organisation, and who reports to whom?
- Are there earn-outs, retention packages or contractor IP to assign?
- Do customer contracts limit hosting, regions or vendors?
A human assessment is Borehole's own paid service, €12,000+ (total price is determined per engagement; travel billed at cost). It is not independent of this report: the same business produced both. It is a separate contract, with its own scope, terms and reliance.
Embed the badge
Tied to this commit and this date, and it fades after 90 days — an honest badge is one that expires. A repository with significant findings gets a neutral "read" badge rather than a red one: the library is where the range is visible, not someone else's README.
[](/r/vllm-project/vllm/f0abb2f)If you embed it: link it to the report, remove it if the report is delisted, and do not present it as a certification or an endorsement.
Rejection log — 92 signals that ran and did not fire
Appendix: rejection log — 92 signals that ran and did not fire
Published on purpose. A signal that looked and decided not to fire has told you something, and this is the calibration data behind every threshold above.
Looked, and found nothing — 66
| Signal | What it found |
|---|---|
| 04.hotspot_concentration | change is spread across the codebase rather than pooling in a few filesfiles 14,498 · top5 share 0.029 |
| 04.debt_markers | debt markers are sparsemarkers 873 · per kloc 0.39 · lines read 2,236,023 |
| 04.churn_trend | rework is not rising against new worklate delete per insert 0.436 · early delete per insert 0.394 |
| 04.fixes_with_tests | fixes usually come with a testfixes 4,767 · share 0.358 · with tests 1,705 |
| 04.fix_share_trend | fixing is not crowding out building |
| 04.compat_layers | little code is kept for the old wayshare 0.001 · code files 3,458 · compat files 2 |
| 06.telemetry_presence | a telemetry or error-reporting library is referencednote presence only; whether anyone watches it is not visible here · matches 59 |
| 06.healthcheck_presence | a health or readiness check is defined |
| 06.telemetry_depth | the service is visible in more than one way |
| 06.print_logging | raw prints are few, or outnumbered by logger callsprints 57 · logger calls 2,366 |
| 06.correlation_ids | a request can be followed through the logstracing True · request id sites 158 |
| 06.environment_branching | environment differences live in configurationsites 0 |
| 06.release_markers | releases are markedtags last year 125 · release commits 2 · changelog commits 5 |
| 06.migrate_on_start | migrations do not run at container start |
| 03.contributor_concentration | no single contributor dominates the last 24 monthsyears 3.7 · window last 24 months · authors 3,506 |
| 03.active_contributors | more than one contributor active in the trailing yearall time 3,506 · active 12mo 2,324 |
| 03.knowledge_silos | every substantial directory has been touched by more than one personauthors 3,506 · directories 10 |
| 03.drive_by_contributors | an open contributor base, where one-commit contributors are normalauthors 3,506 · one off share 0.559 |
| 03.contributor_attrition | no contributor dominates enough for their absence to matterauthors 3,506 · top share less than a quarter · top contributor last committed within the last three months |
| 03.critical_path_ownership | more than one person changes the critical pathsauthors 3,110 |
| 03.second_contributor_attrition | the second and third contributors are still active |
| 03.bus_factor_trend | the team has not narrowed to one person |
| 03.joiners_and_leavers | the team is not shrinkingleft 135 · joined 551 · regular last year 705 |
| 02.merge_discipline | a meaningful share of recent change carries evidence of reviewmerges 0 · window last 24 months · with tracked 6 |
| 02.ci_presence | CI configuration is present in the working treecount 248 |
| 02.test_presence | tests are present in a normal proportion to sourcetests 2,405 · source 5,662 · test ratio 0.425 |
| 02.release_cadence | tags are presenttags 214 |
| 02.conventional_commits | commit subjects generally carry contentsampled 23,617 · low content share 0.001 |
| 02.branch_protection_hints | process scaffolding is present in the repository |
| 02.ci_runs_tests | CI invokes a test runnerconfigs 236 · with tests 64 |
| 02.skipped_tests | few tests are switched offfocused 0 · skipped 36 · test files 2,682 |
| 02.ci_test_gating | no test run in CI is written so that it cannot failconfigs 248 |
| 02.revert_share | change is rarely undoneshare 0.006 · reverts 119 · recent commits 20,531 |
| 02.dependency_update_cadence | dependencies move with the codelockfiles 1 · commits since 40 · days since lockfile changed 0 |
| 02.lint_in_build | a linter or formatter runs in the buildwired 32 · ci configs 248 · product files 3,458 |
| 02.commit_size | changes arrive in reviewable sizesshare 0.019 · commits 19,183 · over 2000 lines 358 |
| 01.burst_import | no commit run met both the rate and the net-addition thresholdgap seconds 180 · min add ratio 3.0 · min net added 2,000 |
| 01.initial_dump | first commit is not a disproportionate share of the codebaseinsertions 1 · share of insertions 0.0 |
| 01.message_uniformity | commit subjects vary in wording and lengthlength stdev 18.82 · repeat share 0.001 |
| 01.revision_depth | most files were revisited after they were first writtenfiles seen 4,212 · write once share 0.377 |
| 01.swallowed_errors | errors are rarely thrown awayloc 2,242,337 · files 52 · per kloc 0.03 |
| 01.build_reproducibility | every base image names a versionfloating 0 · from lines 16 |
| 01.unfinished_migration | no language migration is stalled halfway |
| 01.weights_in_git | no model weights are committedfiles 0 |
| 05.committed_secrets | no credential-shaped strings outside test and fixture pathsnote the working tree only; the history is 05.5 · patterns 8 · files read 6,620 |
| 05.security_guardrails | at least one standard guard rail is configured |
| 05.license_present | a licence file is presentfound True |
| 05.container_runs_as_root | every shipped Dockerfile drops to a non-root user or builds on a non-root base imagedockerfiles 1 |
| 05.secrets_management_pattern | no environment file is committed outside tests and examplesenv files 0 |
| 05.sql_from_strings | no SQL is assembled from strings in product codefiles 0 · sites 0 · value interpolated 0 |
| 05.tls_verification_off | no TLS verification is switched off in product codefiles 0 · sites 0 · pinned instead 0 |
| 05.privileged_workloads | no workload runs privileged or in the host's namespacessites 0 · behind a setting or built here 0 |
| 05.credentials_in_deploy_config | deployment configuration references secrets, never holds themfiles 0 · sites 0 |
| 05.signing_material | no signing keystore or profile is committedfiles 0 |
| 05.dataset_provenance | little data is committeddata files 0 · licence or card files 0 |
| 05.license_copyleft_inherited | every licence found in vendored code is permissive or weak-copyleftvendored paths 43 |
| 05.license_manifest_conflict | manifest and licence file agree on Apache-2.0license Apache-2.0 |
| 05.license_vendored_unlicensed | all 6 vendored components state their terms, in a licence file, their own headers or a notices filecomponents 6 · stated in headers or notices 5 |
| 05.license_grant_unclear | the licence is recognisable as Apache-2.0class permissive · license Apache-2.0 |
| 05.committed_native_libraries | no widely used compiled library is committed to the productlibraries 0 |
| 04.repository_status | the README's first screen states no end, move or freezestatements 0 |
| 02.development_elsewhere | nothing says development happens elsewhereno pr statements 0 · mirror statements 0 · replayed commit share 0.0 |
| 05.licence_changes | the root licence has not changed in the history readlicence versions 1 · unreleased changes 0 |
| 05.paid_edition | no directory carries a separate paid-edition licencedirectories 0 |
| 04.releases_stopped | releases keep pace with developmentmedian gap days 0.0 · days since last tag 4 · commits last 90 days 4,573 |
| 03.single_maintainer | no file says one person maintains itstatements 0 |
Could not conclude, or did not apply — 26
Not a clean result. Each says why it could not answer here.
| Signal | Why |
|---|---|
| 04.dependency_sprawl | no dependency manifest of the product's own was found |
| 04.test_growth | tests are kept inside the files they test, which a count by file name cannot see |
| 04.eol_runtime | no pinned runtime version to check |
| 04.schema_churn | no schema migrations to read |
| 04.dependency_redundancy | no JavaScript manifest to read |
| 06.alert_rules | no alert rules in the repository; they may live in a console |
| 06.deployed_image_floats | no deployment descriptor names an image |
| 06.probes_identical | no Kubernetes probes |
| 06.single_replica | no Kubernetes deployments |
| 06.feature_flags | no feature-flag service; flags may live elsewhere |
| 02.long_lived_branches | too few merges to judge (squash and rebase leave none) |
| 01.notebook_outputs | too few notebooks |
| 05.secrets_in_history | no credential-shaped string was added outside test and fixture paths in the 20,000 most recent commits scanned; older history was not read |
| 05.migration_reversibility | too few migrations to judge |
| 05.resource_limits | no Kubernetes workload manifests |
| 05.public_env_secret | not a frontend |
| 05.unsafe_html | not a frontend |
| 05.source_maps_shipped | not a frontend |
| 05.cleartext_traffic | not a mobile app |
| 04.flags_accumulating | few or no feature flags in the history |
| 04.api_versions_coexisting | the history scanned covers 13 months, less than the 18 this question needs |
| 04.deprecations_lingering | the history scanned covers 13 months, less than the 24 this question needs |
| 04.dependency_churn | the history scanned covers 13 months, less than the 24 this question needs |
| 04.ci_platforms_dropped | the history scanned covers 13 months, less than the 24 this question needs |
| 05.project_advisories | record incomplete: no advisory found in the sources that answered; OSV crates.io vllm-proto (an empty record for a package published here) did not; NVD: 8 records name vllm with nothing tying them to this repository (CVE-2024-11041, CVE-2024-8939, CVE-2024-9053, …); 53 are fixed in the surveyed commit, and the release users run is not placed by this history |
| 05.project_exploited | no affecting advisory is on CISA's Known Exploited list |
Anyone can check this report against the authoritative record — click the stamp, or enter 0S2Y-RN06-9R3T at borehole.dev/verify.
The code commits to the 106 checks this run performed, the bands they produced, and the commit they were read at. It does not cover checks written since.
Put the mark in your README
The image is wrapped in a link to the verification page, so a reader can check it rather than take it on trust. If you embed it: link it to the report, remove it if the report is delisted, and do not present it as a certification or an endorsement.
[](https://borehole.dev/verify/0S2YRN069R3T?via=badge)
5 pieces of evidence on this report quote the repository directly. Four checks quote text: up to five TODO-style comments, up to four commit subjects, up to five unpinned dependency lines and one status line from the README's first screen, such as “no longer maintained”. In each case the text is the evidence, which is why it is kept by default.
This rewrites every stored reading of this repository, and every later reading, including a re-read, is stored without the quoted text. The facts we keep to re-read it still hold the lines, as the privacy notice says. The pointers stay.
Where this scan ran
This scan ran on ordinary cloud infrastructure (cloud-run), not in a hardware-isolated confidential VM. Nothing about it is attested by Google.
What an attested survey proves, and what it does not.
What this scan did with the repository — 7 steps
Written by the steps themselves as they ran, not described afterwards. A hand-written account of a data flow drifts the first time somebody changes the flow and not the account.
| At | Step | Detail |
|---|---|---|
| 0.0s | read repository metadata from the GitHub API | size mb: 317.1authenticated: False |
| 28.1s | cloned the repository to a temporary directory | depth: full |
| 56.8s | read the git history | commits: 23618tracked files: 8367 |
| 485.6s | ran the deterministic checks | blame pass: Falseread working tree: False |
| 485.6s | recorded where this scan ran | attested: Falseenvironment: cloud-run |
| 790.9s | kept the facts, never the source, for re-reading under new checks | |
| 790.9s | deleting the working copy |
Total 790.9s ·source content sent off this machine: none · calls to a language model: 0 · the clone was deleted when the survey ended
An automated reading of a repository's code and its history up to one commit, with public advisory records. It may be wrong, it is not legal, financial, investment or security advice, and it does not replace diligence by a qualified person. It is provided for information: nobody may rely on it (Borehole's terms, section 3).