Install verification: runtime verified (L5) · tested on dsh 0.2.1-rc.2 · last checked Install verification details →
What it doesAI
Model menu for the dsh web GUI that lets you pick the main model, its reasoning effort, and your subagents' models from the composer.
- Three-zone trigger button splits model, effort and subagent choices into three popups
- Search-first model picker with provider column, price badges and list browsing
- Supports assigning a separate model to subagents for multi-model workflows
Capability tags are AI-inferred or rule-extracted from the README (see badge) — vocabulary-limited, not install-verified. AI = inferred; README = literal keyword from the readme.
Installation
dsh plugin --profile web add github:taxueseek/dsh-models#v0.1.4Install method: GitHub · 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 | dsh 0.2.1-alpha.4 |
|---|---|---|
| v0.1.4 | ✓ | · |
| v0.1.0 | · | ✓ |
v0.1.4github✓ 1 runtime-verified2026-10-20
https://github.com/taxueseek/dsh-models
✓dsh 0.2.1-rc.2current2026-10-20L5 · runtime verified
- • Node.js: v24.14.0
- • DSH: 0.2.1-rc.2
- • Plugin artifact: github · https://github.com/taxueseek/dsh-models#v0.1.4
ID: dsh-models@v0.1.4@dsh0.2.1-rc.2 · profile: l4-sandbox
Issued: 2026-10-20 · expires: 2026-10-27
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.
v0.1.0github✓ 1 runtime-verified2026-10-16
https://github.com/taxueseek/dsh-models
✓dsh 0.2.1-alpha.42026-10-16L5 · runtime verified
- • Node.js: v24.14.0
- • DSH: 0.2.1-alpha.4
- • Plugin artifact: github · https://github.com/taxueseek/dsh-models#v0.1.0
ID: dsh-models@v0.1.0@dsh0.2.1-alpha.4 · profile: l4-sandbox
Issued: 2026-10-16 · expires: 2026-10-23
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.
Author badges
Embed the official badges in your README to showcase security & install-test status:
`[](https://www.dsh.so/artifact/dsh-models/)``[](https://www.dsh.so/artifact/dsh-models/)`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 medium, dynamic low — stricter wins → overall medium.
v0.1.4dsh manifestv0.1.4git2026-10-21v0.1.4same versionsuspicious · gitVet is an AST static scan (npm artifact or git source).
show static findings (7)
- warningshellExec · shell
lib/host.js:84execFile(bin, args, { timeout: DIM_TIMEOUT_MS, maxBuffer: DIM_MAX_BUFFER }, (error, stdout) => { - warningeval · code-exec
scripts/interaction.mjs:115eval(readFileSync(join(root, 'lib/client.js'), 'utf8')) - warningnetIp · network
scripts/interaction.mjs:62const dom = new JSDOM('<!doctype html><html><head></head><body><div id="app"></div></body></html>', { url: 'http://127.0.0.1:3080/', pretend - warningeval · code-exec
scripts/perf.mjs:118eval(readFileSync(join(root, 'lib/client.js'), 'utf8')) - warningnetIp · network
scripts/perf.mjs:35const dom = new JSDOM('<!doctype html><html><head></head><body><div id="app"></div></body></html>', { url: 'http://127.0.0.1:3080/', pretend - warningeval · code-exec
scripts/render-count.mjs:65eval(readFileSync(join(root, 'lib/client.js'), 'utf8')) - warningnetIp · network
scripts/render-count.mjs:26const dom = new JSDOM('<!doctype html><html><body><div id="app"></div></body></html>', { url: 'http://127.0.0.1:3080/', pretendToBeVisual: t
show 2 audit findings
- lowSubagent route accepts unauthenticated POST/DELETE and stores caller-supplied provider/model per sessionId
lib/host.js:375The /api/model-picker/subagent handler takes sessionId from the query string and provider/model straight from the request body (the only check is the truthiness test `if (!provider || !model)`), then stores them and the `agent/request` waterfall uses them to override every subagent request for that session. There is no origin, method-token or caller check, and a cross-origin page can send the body as text/plain without a preflight; exploitation still requires knowing the target session id (a UUID), and the chosen provider must already be configured on the host, so impact is limited to silently
- low`dim` CLI is also resolved from PATH as a fallback and executed with host privileges
lib/host.js:78runDim() passes these candidates to execFile (line 84) with fixed arguments, so there is no shell injection and no user-controlled argv. The pinned ~/.local/bin path is tried first, but if it is missing an executable named `dim` earlier in the host's PATH is spawned on every catalog refresh (~10 min) and its JSON output feeds the price/context/discovery data the picker shows. Preconditions are strong: an attacker must already be able to write an executable into a directory on the host PATH, i.e. already have code execution, so this is a hardening gap rather than a live attack path.
Benign model-picker UI plugin. The host half spawns the fixed `dim` CLI and exposes two loopback routes; no secret or code-exec sink from user input. Two low-severity hardening gaps.
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 2026-10-17 · activity: Active
• Repo created: 2026-10-15
• Stars: ★ 1 · Forks: 0
• Health: Active — committed within last 30 days
Source
GitHub: github.com/taxueseek/dsh-models
Author README
View on GitHub ↗The author's original README from the repository — fetched live, unedited (the English page loads the default README).
