wechat-miniapp-radar puts WeChat framework selection, resource browsing and migration questions in one catalogue. The project README describes filters, comparison views and recommendations with evidence sources. It suits teams deciding which framework or library deserves closer investigation. Our review adds source-build measurements at commit 45c6c43, a failed deployment prerequisite and corrections to published descriptions. The useful decision is whether those documented outputs match your team's next task.
What you can do with wechat-miniapp-radar
The README names six features: Radar, Quick Search, Compare, Advisor, Doctor and Weekly. Radar filters resources by recommendation status, risk, type, category and use case; Quick Search finds resources and links to common pages. Compare's documented candidates include Taro, uni-app and native mini-program development. Together, these features organize the selection work that comes before implementing an application with a chosen framework.
Advisor's advertised answer includes a recommendation, conditions for and against the choice, migration cost, next steps and evidence sources. Weekly presents ecosystem reports and recent risk signals. The README's feature descriptions give a team several possible outputs to request: a comparison, a reasoned recommendation or an ecosystem update. Choose the output that matches the decision at hand, then follow its framework and library references for the implementation details.
What people use wechat-miniapp-radar for
Issue #187 records steedos proposing a mini-program resource and later reporting a pull request after justjavac requested one. The resource-submission exchange belongs to the repository's earlier resource-list incarnation. It documents a contribution workflow rather than a current application's customer result: a contributor brought a project to the maintainer, the maintainer asked for a pull request, and the contributor said it had been submitted.
The current README names product, engineering and architecture teams making technology choices, teams assessing frameworks and surrounding libraries, and resource-list maintainers. These are documented intended uses, with selection teams seeking candidates and maintainers organizing a collection. For either group, the catalogue's categories offer a starting structure: official documentation, tools, plugins, components, backend SDK components and demos. The historical submission shows how one resource entered a discussion about that collection.
What happened when we ran it
Our November 18, 2026 run at commit 45c6c43 installed in 60 seconds and built in 28 seconds. The source-run measurements recorded 384 dependencies and 617 MB installed, against a 1.6 MB checkout with 177 files and approximately 17,381 source lines. The recorded sandbox configuration was 3 CPUs, 8 GB RAM and image label lab-node:22. For storage planning, the installed footprint was substantially larger than the checkout.
| Recorded check | Result | What the result describes |
|---|---|---|
| npm install | Succeeded in 60 seconds | Source installation at commit 45c6c43 |
| Build | Succeeded in 28 seconds | Compilation at the measured commit |
| Dependencies and disk | 384 dependencies; 617 MB installed | Installed dependency footprint |
| Tests | No tests script/target; skipped | The harness's generic test step |
| npm audit | 26 vulnerabilities: 1 critical, 13 high, 12 moderate, 0 low | Dependency advisory totals |
| Repository scan | 177 files; approximately 17,381 source lines; 1.6 MB checkout | Source size |
| Automation and packaging scan | 4 CI workflow files; no Dockerfile; no tests directory | File-presence observations |
The November 19 attempt failed at the runner's Compose prerequisite. Its safety-step log reads unsafe repo: none of compose.yaml, compose.yml, docker-compose.yaml, docker-compose.yml exists (rule: a compose file is required). Cleanup subsequently returned teardown: ok in the teardown record. This is a specific account of where that attempt stopped; the earlier source installation and build retain their recorded successful results.
The current package manifest includes check, doctor:test, advisor:test and security:test. Its checking code therefore deserves a more precise description than "no tests": the harness skipped its generic test step, while the manifest lists named checking commands. The table preserves the actual harness outcome rather than replacing it with a count derived from script names.
Where people get stuck: support questions and metadata errors
Issue #4 reports a failed WeChat quickstart after the reporter created a project without an appId. The developer-tool question names touristappid, version v.09.092300 and Mac OSX. Replies suggested supplying an appId or disabling the developer tool's proxy. That environment identifies the subject of the report: WeChat developer tooling. Other historical threads concern component behavior, resource availability and dates on news items.
| Report or analysis | Reported behavior | Response or proposed remedy |
|---|---|---|
| Issue #36, picker component | Switching selections produced Unexpected token u in JSON at position 0 |
The synthesis records a referral to a WeChat support group |
| Issue #81, unavailable forum | The recommended weiapp.online forum returned Server not found |
raymondyan acknowledged its disappearance |
| Issue #83, news dates | fuchao2012 requested publication dates because news ages | justjavac welcomed the suggestion |
| Standard Beagle, enrichment | GitHub or npm metadata requests can return non-OK responses | The analysis recommends correcting URLs and retrying transient failures |
| Standard Beagle, Upstash | An HTTP request can succeed while the response body rejects the command | The analysis recommends inspecting the error and catching it with an in-memory fallback |
Standard Beagle describes the enrichment collector converting fetch failures into a CollectedSignal with ok:false and an error message. Its enrichment analysis recommends correcting or removing resource URLs for 404 responses and regenerating data. The Upstash analysis addresses a different error path: a successful HTTP response whose body rejects the command. These are attributed code analyses and proposed remedies, rather than incident reports from application users.
Repository activity and resource upkeep
GitHub's November 19 observation reports a last push of November 11, 2026 and zero open issues and pull requests. The repository metadata identifies TypeScript as the primary language. Those dated fields help a reader locate the checkout in the repository's history. The recorded build uses commit 45c6c43, whose commit date is also November 11, 2026.
The historical forum disappearance in issue #81 and request for news dates in issue #83 point to concrete catalogue-maintenance tasks: removing dead destinations and giving readers publication dates. A maintainer evaluating Radar can use those tasks to frame an initial review of imported resources. They concern the links and information readers encounter, beyond the application files tracked by GitHub.
Self-hosting starts with Node.js 20 and the manifest
The package manifest requires Node.js 20 or newer and maps dev, build and start to Next.js commands. It also defines database generation, migration, push and import operations. The manifest's script definitions are the concrete command reference for someone reading the application's setup. They expose both application startup and database work, which belong in the operating plan for a separate instance.
GitHub Locker describes self-hosting as requiring Node.js 20+ and PostgreSQL, with Drizzle ORM over Postgres. That deployment summary is consistent with the manifest listing drizzle-orm and postgres dependencies. The primary README leads with the hosted site, audiences and feature routes. A hosting decision therefore starts with a choice between using that advertised service and taking responsibility for an application with database commands and dependencies.
Chinese documentation and GPLv3 shape reuse
The README is Chinese, and the repository's LICENSE contains GNU General Public License version 3. The README also names QQ discussion groups, including one marked full. For an English-speaking team, reading those product descriptions and historical discussions requires Chinese comprehension or translation. The implementation's TypeScript language is a separate practical consideration for whoever will maintain the application code.
The GPLv3 text permits charging for conveyed copies and sets conditions for conveying source and modified works. It also says interaction through a computer network without transferring a copy is not conveying. A commercial-use review should identify the proposed activity, such as internal operation or distribution of a modified copy, and apply the relevant licence provisions to that activity.
Security starts with the 26 dependency advisories
npm audit reported 26 vulnerabilities: 1 critical, 13 high and 12 moderate. The audit result from our source run makes dependency triage a concrete hosting task at commit 45c6c43. Inspecting affected dependency paths and the effects of upgrades is the useful next step for an operator. The audit totals stay attached to the measured checkout, alongside its successful installation and compilation.
The GitHub security page reports that the project has no SECURITY.md policy. AI Slop Code Risk Monitor, meanwhile, describes its report signals as prompts for human review. Its report explanation directs reviewers toward questions about code intent, dependencies and scaffolding. For an operator, the immediately actionable evidence here is the dependency advisory count and the absence of a documented security policy.
wechat-miniapp-radar vs alternatives
A static awesome list addresses resource discovery most directly; Renovate and Dependabot-Core address dependency maintenance after selection. Radar's README describes adding filters, comparisons and recommendation outputs to a resource collection. That makes a static list the closer comparison for readers deciding how much selection machinery they need. Dependency tools become relevant when the desired output is a version change in an existing project.
| Option | Documented job | Choose it when |
|---|---|---|
| wechat-miniapp-radar | Resource filters, framework comparisons and selection advice | You want to organize a WeChat stack decision |
| A static awesome list | A collection of resource links | Browsing links meets the immediate need |
| Renovate | Pull requests updating dependency versions | You want ongoing version-update proposals |
| Dependabot-Core | Dependency-update logic, including manifests and lockfiles | You are building dependency-maintenance automation |
Renovate creates pull requests to update dependency versions. Its README describes discovering dependency references and proposing newer versions. Dependabot-Core generates updated manifests and lockfiles for a new dependency version. The Dependabot-Core README identifies it as a library and recommends Dependabot CLI for standalone use; that CLI creates dependency diffs and requires additional wiring to turn them into pull requests.
Outdated claims, corrected
SourcePulse describes Doctor as scanning local mini-program projects, whereas the README specifies pasted project configuration. The SourcePulse description broadens the input beyond the primary feature description. The latter says Doctor identifies framework dependencies, outdated approaches and migration risks from that configuration. Read its output as advice about the declarations entered, rather than a report on source files throughout a repository.
The README advertises 237 resources, but its categories total 235: 6 official documents, 44 tools, 13 plugins, 64 components, 2 backend SDK components and 106 demos. The README's data overview contains both the headline and breakdown. GitHub Locker has another mismatch: its headline is 236, while 106 demos, 64 components, 43 tools, 13 plugins, 6 official documents and 2 backend SDK components total 234.
Hysen Labs' maintenance discussion gives an October 2 last-push date, while its November 18 snapshot says commits arrived eight days earlier. Those conflicting maintenance statements should be replaced with the dated GitHub observation: November 11, 2026. For this review, that is the repository-activity reference; the catalogue arithmetic above is reported directly from each document's own figures.
FAQ
The README links miniapp.denohub.com as the project's main site, so installation is a choice for readers who want their own instance. The hosted entry point answers the browsing question first. The following questions cover that access choice, local requirements and Doctor's input, each raised in published questions about the project. Detailed commands and corrections appear in their respective sections above.
Can I use the hosted website, or do I need to install it?
The README directs readers to miniapp.denohub.com and lists routes for Radar, Compare, Advisor, Doctor and Weekly. That is the advertised route for using the product without operating a separate copy. Self-hosting means taking on the application's startup and database work; it is relevant when running your own instance is itself a requirement, rather than merely browsing the catalogue.
How do I run wechat-miniapp-radar locally, and what services does it require?
Node.js 20+ is the manifest's runtime requirement, and GitHub Locker describes PostgreSQL for a separate instance. The package manifest contains the application and database command reference discussed in the self-hosting section. This gives a prospective operator specific files and operations to examine before choosing a hosting arrangement, rather than treating the README's main-site link as deployment instructions.
How does Doctor check a project, and what input does it accept?
Doctor accepts pasted project configuration, according to the README. Prepare the relevant configuration for a migration discussion; use Advisor when the starting input is a technology-selection question instead. This input choice determines which feature to open. The corrections section explains why the broader description of scanning local projects gives readers the wrong expectation about what they will submit.
What we have not tested yet
commit 45c6c43 has installation and build results, but no completed application workload or recommendation-accuracy result. Advisor and Doctor accuracy, hosted browsing, database migration, scheduled refreshes, English UI behavior, commercial deployment and load behavior are outside the executed checks. The manifest's named checking commands have no recorded execution results. Actual runtime and operating-system versions require raw toolchain logs; lab-node:22 is only the recorded image label.
The November 19 run header says deployment used the project's own Docker Compose with ports bound to 127.0.0.1, but the safety step found no Compose file. Successful deployment logs, completed hands-on Q&A and review screenshot records are absent. The runner's unsafe repo label is not a software-security finding, and aggregate audit totals neither identify affected packages nor demonstrate an exploit.
Standard Beagle's extracts give no supported analysis date or commit identifier, leaving applicability to 45c6c43 unresolved. Recovery after enrichment failure, catalogue freshness and the proposed memory fallback's persistence and consistency need separate checks. Historical replies do not confirm technical resolutions, issue closure does not show implementation, and issue #187 does not show acceptance. Unrelated social and video results add no adoption evidence; repository activity and pattern scores establish neither support quality nor security or AI authorship.


