bytedance/UI-TARS-desktop

Public scan — anyone with this URL can view this analysis. Sign up to track your own repos privately, run scheduled re-scans, and get AI fix prompts via your dashboard.

https://github.com/bytedance/UI-TARS-desktop · scanned 2026-05-15 21:14 UTC (2 weeks, 6 days ago) · 10 languages

527 findings (15 legacy + 512 scanner) 31st percentile · Typescript · large (100-500K LoC) Scanner says 46 (higher by 25)

File as issue Image v2 schema

UNIFIED Repobility · multi-layer engine · AI coders

Complete repo analysis

Last scanned 2 weeks, 6 days ago · v1 · 11 findings from 1 source. Findings combine the legacy security pipeline AND the multi-layer engine (atlas, wiring, flows, ranked) AND verified AI agent contributions.

JSON

Score breakdown â 2026-05-17-v4 calibration-aware

Component	Sub-score	Weight	Contribution
`structure_score`	60.0	0.15	9.00
`security_score`	86.2	0.25	21.55
`testing_score`	49.0	0.20	9.80
`documentation_score`	75.0	0.15	11.25
`practices_score`	75.0	0.15	11.25
`code_quality`	80.0	0.10	8.00
Overall		1.00	70.8

Calibrated penalty buckets (security_score): docker: 2.6 Â· threat: 11.2

Severity distribution — click a segment to filter

Active filters: excluding tests × Reset all

Severity: Critical 0 High 1 Medium 4 Low 5 9-Layer: Software Security Quality Integrity Frontend Hardware Data Network Cicd Source: Legacy 11 9-layer 0 Crowd 0 Layer: Cicd 2 Security 6 Quality 3

Exclude dismissed / FP Exclude test files

Scan summary Repository scanned at 45.6/100 with 100.0% coverage. It contains 5243 nodes across 30 cross-layer flows, written primarily in mixed languages. Engine surfaced 0 findings. Risk profile is low: 0 critical, 0 high, 0 medium. Recommended next step: open the software layer findings first — that's where the highest-impact wins live.

Showing 10 of 11 findings. Click TP / FP to vote on a finding's accuracy — votes adjust the confidence weighting and improve detection across the platform.

medium Legacy quality error_handling conf 1.00 [ERR002] Empty Catch Block: Empty catch blocks hide errors.

Log the error or rethrow it. Use console.error() at minimum.

packages/agent-infra/browser/src/browser-finder/firefox-paths.ts:19 error_handlinglegacy

medium Legacy quality error_handling conf 1.00 [ERR002] Empty Catch Block: Empty catch blocks hide errors.

Log the error or rethrow it. Use console.error() at minimum.

packages/agent-infra/browser/src/browser-finder/chrome-paths.ts:40 error_handlinglegacy

medium Legacy quality error_handling conf 1.00 [ERR002] Empty Catch Block: Empty catch blocks hide errors.

Log the error or rethrow it. Use console.error() at minimum.

apps/ui-tars/src/renderer/src/pages/remote/cast/canvas.tsx:242 error_handlinglegacy

low Legacy security deserialization conf 1.00 [SEC007] Unsafe Deserialization: Unsafe deserialization can execute arbitrary code.

Use yaml.safe_load() instead of yaml.load(). Avoid pickle for untrusted data.

packages/ui-tars/cli/src/cli/start.ts:40 deserializationlegacy

low Legacy security deserialization conf 1.00 [SEC007] Unsafe Deserialization: Unsafe deserialization can execute arbitrary code.

Use yaml.safe_load() instead of yaml.load(). Avoid pickle for untrusted data.

scripts/merge-yml/merge-yml.ts:25 deserializationlegacy

low Legacy security deserialization conf 1.00 [SEC007] Unsafe Deserialization: Unsafe deserialization can execute arbitrary code.

Use yaml.safe_load() instead of yaml.load(). Avoid pickle for untrusted data.

apps/ui-tars/src/main/store/setting.ts:100 deserializationlegacy

low Legacy security llm_injection conf 0.80 [SEC017] Unbounded Input to LLM/External API: User input is passed to an LLM or external AI API (OpenAI, Anthropic, etc.) without any visible length or size validation. This creates two risks: (1) Cost abuse — an attacker can send extremely long inputs to burn through your API credits (a single 128K-token request to GPT-4 costs ~$4, and automated attacks can drain budgets in minutes). (2) Context stuffing — oversized inputs can push your system prompt out of the context window, effectively disab

1) Enforce a maximum input length BEFORE sending to the API: e.g. `if len(text) > 4000: return error`. 2) Use token counting (tiktoken for OpenAI, anthropic's token counter) to enforce token-level limits. 3) Set max_tokens on the API call to cap response cost. 4) Add rate limiting per user/IP to pr…

packages/agent-infra/browser-use/src/agent/agents/navigator.ts:97 llm_injectionlegacy

packages/agent-infra/browser-use/src/agent/agents/base.ts:142 llm_injectionlegacy

medium Legacy cicd docker conf 0.90 Docker build context has no .dockerignore

Without .dockerignore, build context can include source history, local env files, dependencies, and generated artifacts.

.dockerignore dockerlegacy

high Legacy cicd docker conf 0.82 Docker final stage has no non-root USER

Docker images run as root unless the image or Dockerfile switches to a non-root user.

packages/agent-infra/mcp-servers/filesystem/Dockerfile:13 dockerlegacy

For AI agents: Voting guide (TP/FP) MCP manifest Stdio wrapper SARIF Integrate Findings queue Vote TP/FP on findings to calibrate the engine.

For AI agents + API integrations

Email me when this repo regresses

Free. We re-scan periodically; new criticals → your inbox. No signup required for the scan itself.

API access

This page is publicly accessible at: https://repobility.com/scan/b8acb5fe-7d86-484e-ad15-3a3f42605a19/

To check status programmatically (no auth required):

curl -s https://repobility.com/api/v1/public/scan/b8acb5fe-7d86-484e-ad15-3a3f42605a19/

Important — please don't re-submit the same URL repeatedly. The submission endpoint is idempotent: re-submitting the same git URL returns this same scan_token, not a new one. To re-scan this repo, sign up free and use the dashboard.