中文 EN

Scan Report

Scan New Files

✓ Trusted — Risk Score 5/100
Last scan:2 days ago Rescan
5 /100
tx-cos
面向 OpenClaw 的腾讯云 COS 原生操作技能
tx-cos is a legitimate Tencent Cloud COS object storage skill using only the official cos-nodejs-sdk-v5 SDK for cloud operations, with no malicious behavior, exfiltration, or hidden functionality.
Skill Nametx-cos
Duration38.7s
Enginepi
✓
Safe to install
No action needed. Consider adding an explicit allowed-tools declaration in SKILL.md for transparency.

Findings 2 items

Severity Finding Location
Low
Missing allowed-tools resource declaration
SKILL.md lacks an explicit allowed-tools section mapping node/npm to resources (filesystem:READ for file uploads, environment:READ for credentials, network:READ for COS API calls). This is a documentation transparency gap, not a security violation.
metadata.openclaw.requires.bins only declares node/npm without resource-level mapping
→ Add an allowed-tools declaration: filesystem:READ (for local file reads during upload), environment:READ (for COS_SECRET_* vars), network:READ (for Tencent Cloud API calls)
 SKILL.md:1
Low
Credential access without exfiltration
Scripts read COS_SECRET_ID and COS_SECRET_KEY from environment variables but only pass them to the official Tencent Cloud SDK for authentication. Credentials are never logged, echoed, or transmitted outside the SDK's encrypted COS API calls.
SecretId: process.env.COS_SECRET_ID, SecretKey: process.env.COS_SECRET_KEY
→ No action needed — this is expected and necessary behavior for the skill's core function
 scripts/lib.js:38
ResourceDeclaredInferredStatusEvidence
Filesystem NONE READ ✓ Aligned scripts/upload_file.js:16, scripts/upload_batch.js:11 — reads local files via fs…
Environment NONE READ ✓ Aligned scripts/lib.js:32 — accesses COS_SECRET_ID, COS_SECRET_KEY, COS_REGION, COS_BUCK…
Network NONE READ ✓ Aligned scripts/lib.js:38 — instantiates cos-nodejs-sdk-v5 for Tencent Cloud COS API cal…
Shell NONE NONE — No child_process, exec, or shell invocation found across all scripts
Skill Invoke NONE NONE — No skill self-invocation or cross-skill calls
Clipboard NONE NONE — No clipboard access
Browser NONE NONE — No browser automation
Database NONE NONE — No database access

File Tree

11 files · 21.0 KB · 739 lines
JavaScript 10f · 566L Markdown 1f · 173L
├─ ▾ 📁 scripts
│ ├─ 📜 check_config.js JavaScript 24L · 594 B
│ ├─ 📜 delete_folder.js JavaScript 53L · 1.7 KB
│ ├─ 📜 generate_signed_url.js JavaScript 32L · 774 B
│ ├─ 📜 get_folder_stats.js JavaScript 49L · 1.5 KB
│ ├─ 📜 lib.js JavaScript 101L · 2.3 KB
│ ├─ 📜 list_objects.js JavaScript 38L · 964 B
│ ├─ 📜 manage_objects.js JavaScript 85L · 2.8 KB
│ ├─ 📜 upload_batch.js JavaScript 86L · 2.3 KB
│ ├─ 📜 upload_file.js JavaScript 44L · 994 B
│ └─ 📜 upload_large_file.js JavaScript 54L · 1.5 KB
└─ 📝 SKILL.md Markdown 173L · 5.7 KB

Dependencies 1 items

PackageVersionSourceKnown VulnsNotes
cos-nodejs-sdk-v5 * npm No Official Tencent Cloud COS SDK; version not pinned but this is a standard SDK dependency

Security Positives

✓ All 10 scripts are pure Node.js with no shell, exec, or child_process calls
✓ Uses only the official cos-nodejs-sdk-v5 SDK — no curl/wget, no eval, no base64 decoding
✓ No credential exfiltration: environment variables used only for SDK auth
✓ No access to sensitive local paths (~/.ssh, ~/.aws, .env files)
✓ No remote code execution, reverse shells, or C2 indicators
✓ No hidden functionality in HTML comments or whitespace
✓ Output sanitization: SKILL.md explicitly requires no credential leakage in responses
✓ Deletion operations have confirmatory logic with user-facing safeguards
✓ No dependency installation at runtime (pip/npm install not present)