Install verification: runtime verified (L5) · tested on dsh 0.2.1-rc.2 · last checked Install verification details →
What it does
DSH 插件:每天替你刷一轮 B站,读字幕挑出值得的、写一份日报,值得的 @ 你
Installation
dsh plugin --profile web add dsh-bili-scrollInstall method: npm · runtime verified (L5)
Verification & Compatibility
Grouped by plugin version: each card lists the dsh versions actually tested against that artifact version and their install verdicts (L5 wins). Untested combinations are omitted — absence means untested, never assumed compatible. L1–L3 are a plugin-level static archive (independent of the dsh version — the same verdict on every dsh row): they sit in each row's level grid next to L4/L5, marked "static", while their evidence is stored once per plugin inside that row's "checks & evidence" instead of being duplicated per version.
External credentials (plugin-level): none detected
○ The author declares no dsh version requirement — install is not blocked by version.
| plugin \ dsh | dsh 0.2.1-rc.2 current |
|---|---|
| 0.1.7 | ✓ |
0.1.7npm✓ 1 runtime-verified2026-10-21
dsh-bili-scroll
✓dsh 0.2.1-rc.2current2026-10-21L5 · runtime verified
- • Node.js: v24.14.0
- • DSH: 0.2.1-rc.2
- • Plugin artifact: npm · dsh-bili-scroll@0.1.7
ID: dsh-bili-scroll@0.1.7@dsh0.2.1-rc.2 · profile: l3-fullscan
Issued: 2026-10-21 · expires: 2026-10-28
View checks & evidence
active in loader tree (150 entries)
L1–L3 are plugin-level static conclusions (baseline, identical across dsh versions): evidence is stored once per plugin.
Only the latest 3 published versions are shown; source-snapshot (commit hash) groups, unstated groups and older versions are omitted — the full archive stays in the data layer (snapshot/unstated groups appear only when a plugin has no published version).
Author badges
Embed the official badges in your README to showcase security & install-test status:
`[](https://www.dsh.so/artifact/dsh-bili-scroll/)``[](https://www.dsh.so/artifact/dsh-bili-scroll/)`Security Report
Two automated layers, neither a manual review: a static scan (per-plugin rule checks) and a model-driven dynamic security audit that covers only the subset static-flagged at warn or above and never executes plugin code.
Static low, dynamic medium — stricter wins → overall medium.
v0.1.7dsh manifest0.1.7npm2026-10-21v2.0.0version moved onclean · gitVet is an AST static scan (npm artifact or git source).
show static findings (8)
- warningchildProcess · code-exec
index.js:809const child = spawn(dsh.command, args, { - warningchildProcess · code-exec
index.js:931const c = spawn('uv', ['--version'], { windowsHide: true, stdio: 'ignore' }) - warningchildProcess · code-exec
index.js:945const c = spawn(cmd, ['--version'], { windowsHide: true, stdio: 'ignore' }) - warningnetIp · network
index.js:1338try { await handle(req, res, new URL(req.url, 'http://127.0.0.1')) } catch (e) { - warningchildProcess · code-exec
lib/login.mjs:97child = spawn(pythonPath || 'python', ['-c', code, url, png], { cwd: stateDir, windowsHide: true }) - warningchildProcess · code-exec
lib/mcp.mjs:36this.proc = spawn(command, args, { - warningchildProcess · code-exec
lib/uvfetch.mjs:69const r = spawnSync(cmd, args, { stdio: 'ignore' }) - warningchildProcess · code-exec
lib/uvfetch.mjs:147const r = spawnSync(target, ['--version'], { stdio: 'ignore' })
show 5 audit findings
- mediumUnauthenticated /api/bili-duty routes let a caller set pythonPath/dshBin and then execute them
index.jsThe request handler performs no authentication or origin check; it relies entirely on whatever the host webServer applies. A caller that can reach the endpoint can set dshBin (or pythonPath, or install-asr) to any existing file and then POST /api/bili-duty/run or /api/bili-duty/install, which makes the plugin spawn that file via process.execPath or as the MCP command — arbitrary code execution in the user's context. POST /api/bili-duty/run needs no body, so a cross-site form POST can also start a round that posts comments with the user's Bilibili account.
- mediumShell fallback spawn passes config/AGENTS.md-derived prompt text through a shell
index.jsWhen the dsh binary cannot be located (the comments state this happens on Electron desktop installs), resolveDsh returns shell:true and the prompt — which embeds cfg.headlessProfile, cfg.keywords, cfg.mentionName and up to 6000 chars of AGENTS.md read from cfg.workspace — is concatenated into a shell command line. Any ';', '$(...)' or backtick in those values (e.g. a cloned repo's AGENTS.md, or a config value written via the endpoint above) becomes command execution as the DSH user.
- mediumDownloads an executable uv binary from third-party proxies and validates it against a checksum served by the same proxy
lib/uvfetch.mjsuv.exe/uv is downloaded from gh-proxy.com / ghproxy.net and then executed. For those sources the expected sha256 is fetched from the same proxy URL prefix, so a tampering proxy can serve a matching digest and the check still 'passes'; when no digest is returned the code proceeds anyway. Result: an executable from a third-party mirror runs on the user's machine with the plugin's privileges.
- lowpythonPath is interpolated unescaped into the generated headless overlay YAML
index.jsensureOverlay writes a --patch file consumed by dsh. A pythonPath containing a single quote breaks out of the YAML scalar and can inject arbitrary patch entries (e.g. an extra plugin/MCP entry) into the headless round. Reachable only by whoever can set pythonPath (settings page or the HTTP config route).
- lowBilibili session cookies stored as plaintext JSON with default file permissions
lib/login.mjsSESSDATA/bili_jct (full account power, incl. posting) are persisted unencrypted in $DSH_HOME/bili-duty/bili_credential.json with umask-default permissions, readable by any process/user with access to that directory.
Bilibili duty plugin is mostly coherent, but its unauthenticated /api/bili-duty control routes plus shell-fallback spawn and self-updating uv downloader create real medium-risk execution surfaces.
Rule: the dynamic audit can only raise risk, never lower it (raise-only, never an exemption); a wrong raise is withdrawn by human review and the tier falls back to the static verdict. Overall verdict = the stricter of the two layers.
Disclaimer: the static scan and the dynamic security audit are both automated analysis, not a security guarantee. Always review what you install.
Activity
Last commit 2026-10-21 · activity: Active
• Repo created: 2026-10-14
• Stars: ★ 335 · Forks: 43
• Health: Active — committed within last 30 days
Source
Author README
View on GitHub ↗The author's original README from the repository — fetched live, unedited (the English page loads the default README).
