Read 2026-10-23
maximhq/bifrost: technical due diligence, read from its repository
What Borehole read in maximhq/bifrost's code and history at 4810249: its licence, security advisories, releases, review practice, tests, dependencies and operations. Every finding cites the file and line it rests on. The full report also covers the team and asks the questions the code cannot answer.
| Dimension | Rating |
|---|---|
| Architecture & Codebase — including AI provenance | strong |
| Engineering Process & SDLC Maturity | adequate |
| Product & Engineering Maturity | strong |
| Security & Reliability Posture | adequate |
| Observability & Operations | adequate |
| Fit with the Acquirer | not assessable |
Architecture & Codebase — including AI provenance
strongframework/configstore/migrations.go is 16,395 lines, 94× this codebase's median file
adequateThe median source file here is 174 lines. 70 files exceed 1,740 lines, the largest being 16,395. 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.
framework/configstore/migrations.go:1 — 16,395 lines (94× median) |
core/providers/anthropic/responses.go:1 — 11,033 lines (63× median) |
core/bifrost.go:1 — 10,989 lines (63× median) |
transports/bifrost-http/lib/config.go:1 — 10,768 lines (61× median) |
framework/configstore/rdb.go:1 — 10,756 lines (61× median) |
Engineering Process & SDLC Maturity
adequateTests are retried until they pass
adequateThe suite is configured to retry failing tests (tests/e2e/playwright.config.ts). 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.
tests/e2e/playwright.config.ts:69 — test retry |
7% of recent commits change more than 2,000 lines
adequate339 of 5058 commits in the last 24 months changed over 2,000 lines of code, not counting reformatting, renames, vendored or generated code. A change that size is rarely reviewed line by line. Ask how they were reviewed, and what made them that large.
| 339 commits over 2,000 lines |
Product & Engineering Maturity
strongNo finding.
Security & Reliability Posture
adequate4 credential-shaped strings in the tree sit in tests, examples, documentation or local development settings
adequateStrings shaped like credentials appear only where fakes are usual: test fixtures, examples, documentation, placeholders and local stacks. They are almost always fakes. Ask someone to confirm that none of them works anywhere, because a real key in a development file is still a real key.
| 2 cited: AWS access key, where fakes are usual |
| 2 cited: JWT, where fakes are usual |
4 outbound calls can wait forever
adequate4 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.
cli/internal/client/client.go:378 — outbound call without a timeout |
core/agent/manager.go:557 — outbound call without a timeout |
core/providers/utils/utils.go:2095 — outbound call without a timeout |
30 SQL statements built from strings
adequate30 places in product code build SQL by formatting or joining strings rather than passing parameters, and in 7 of them a value, not a table or column name, goes into the string. Where any of those values comes from a request, it is an injection point. Each site needs a look; the fix, parameters, is small.
framework/configstore/migrations.go:5975 — SQL built from a string |
framework/configstore/migrations.go:15633 — SQL built from a string |
framework/queue/clickhousestore.go:288 — SQL built from a string |
2 locked dependencies have published vulnerabilities, 3 high or critical in what ships
adequate2 of 704 locked dependencies match 4 published advisories in OSV, 3 of them rated high or critical. 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.
ui/package-lock.json — dompurify 3.4.16: GHSA-cg2g-xh64-v4q7 (critical), fixed in 3.4.17 |
ui/package-lock.json — zod 4.2.1: GHSA-mrhx-c7xc-m3w5 (high), fixed in 4.6.6 |
5 advisories against this project in the last 24 months, 5 high or critical
adequateThe newest was published on 2026-09-23. 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 2 of them are fixed in the code you ship could not be told from this history. Answer lives in: the release you run, set against each fixed version. The record is incomplete: NVD: 44 records name bifrost with nothing tying them to this repository (CVE-2021-28663, CVE-2021-28664, CVE-2021-29256, …).
| 2026-09-23: GHSA-2qp8-4xgm-fw6g (high), fixed in 2.1.0: Unauthenticated RCE via Custom Plugin HTTP Path on Dynamically Linked Builds |
| 2026-09-23: GHSA-86gf-xh3g-rvxq (critical), fixed in 2.1.0: Bifrost MCP Stdio RCE |
| 2026-09-14: CVE-2026-90898 (critical): Bifrost registers MCP clients through its management API. A stdio client is a command plus args. Bifrost starts that program in the gateway the moment the clien |
| 2026-09-06: CVE-2026-86242 (high): Bifrost HTTP transport before 2.0.0 accepts an enabled custom plugin whose path is an HTTP URL through unauthenticated POST /api/plugins when management authent |
| 2026-07-21: GHSA-w98g-5w9p-p3rc (high), fixed in >=1.5.16: SSRF deny-list incomplete: isPublicIP permits CGNAT, IPv6 6to4/NAT64, and site-local in FetchAndEncodeURL |
Observability & Operations
adequateNo 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.
| no error reporter initialised |
Readiness and liveness probes are identical in 4 of 7 workloads
adequateBoth probes call the same endpoint. When a dependency slows down, the liveness probe fails with the readiness probe, so the platform restarts pods that only needed taking out of rotation, and a partial outage becomes a full one. Liveness should ask whether the process is alive; readiness whether it can serve.
helm-charts/bifrost/templates/qdrant-deployment.yaml — same probe endpoint twice |
helm-charts/bifrost/values-examples/providers-and-virtual-keys.yaml — same probe endpoint twice |
helm-charts/bifrost/values.yaml — same probe endpoint twice |
Fit with the Acquirer
not assessableNot rated: structurally not assessable from a repository.
Read more, or read your own
The full report on maximhq/bifrost: every dimension, the questions for the room, and the audit log.
Survey a repository: free on public repositories. Private code is read on your own machine by a collector whose source you can read first.