Scan Report
5 /100
nano-gpt
NanoGPT API CLI skill for text, image, and video generation
This is a legitimate NanoGPT API CLI wrapper skill with clear documentation and appropriate, declared capabilities. The pre-scan IOCs are false positives: base64 decoding is standard image handling, and test credentials are test fixtures.
Safe to install
No action required. The skill is safe to use as documented.
| Resource | Declared | Inferred | Status | Evidence |
|---|---|---|---|---|
| Filesystem | READ | READ | ✓ Aligned | SKILL.md declares local image/video inputs; code reads only user-specified files… |
| Filesystem | WRITE | WRITE | ✓ Aligned | SKILL.md declares --output flag for images/videos; code writes to user-specified… |
| Network | WRITE | WRITE | ✓ Aligned | SKILL.md declares nano-gpt.com API calls; client.ts makes authenticated POST/GET… |
| Environment | READ | READ | ✓ Aligned | SKILL.md declares NANO_GPT_API_KEY and NANO_GPT_* overrides; config.ts reads onl… |
| Shell | NONE | NONE | — | Scripts exec node binary only; no subprocess with user input |
| Skill Invoke | NONE | NONE | — | No skill invocation found |
| Clipboard | NONE | NONE | — | No clipboard access |
| Browser | NONE | NONE | — | No browser automation |
| Database | NONE | NONE | — | No database access |
1 Critical 1 High 10 findings
Critical Encoded Execution Base64 编码执行（代码混淆）
Buffer.from(image.b64_json, "base64" cli/src/cli.ts:551 High API Key 疑似硬编码凭证
apiKey: "abcdefgh12345678" cli/test/config.test.ts:40 Medium External URL 外部 URL
https://docs.nano-gpt.com/ README.md:7 Medium External URL 外部 URL
https://nano-gpt.com README.md:45 Medium External URL 外部 URL
https://cdn.example/video.mp4 cli/test/cli.test.ts:41 Medium External URL 外部 URL
https://proxy.example/api cli/test/client.test.ts:33 Medium External URL 外部 URL
https://proxy.example/api/v1/models cli/test/client.test.ts:47 Medium External URL 外部 URL
https://file.example cli/test/config.test.ts:14 Medium External URL 外部 URL
https://env.example/ cli/test/config.test.ts:22 Medium External URL 外部 URL
https://env.example cli/test/config.test.ts:35 File Tree
31 files · 74.4 KB · 2756 lines TypeScript 16f · 2261L
Markdown 4f · 294L
JSON 4f · 142L
Shell 6f · 55L
YAML 1f · 4L
├─
▾
agents
│ └─
openai.yaml
YAML
├─
▾
cli
│ ├─
▾
src
│ │ ├─
bin.ts
TypeScript
│ │ ├─
cli.ts
TypeScript
│ │ ├─
client.ts
TypeScript
│ │ ├─
config.ts
TypeScript
│ │ ├─
constants.ts
TypeScript
│ │ ├─
image-input.ts
TypeScript
│ │ ├─
messages.ts
TypeScript
│ │ ├─
stream.ts
TypeScript
│ │ └─
types.ts
TypeScript
│ ├─
▾
test
│ │ ├─
cli.test.ts
TypeScript
│ │ ├─
client.test.ts
TypeScript
│ │ ├─
config.test.ts
TypeScript
│ │ ├─
image-input.test.ts
TypeScript
│ │ ├─
messages.test.ts
TypeScript
│ │ ├─
stream.test.ts
TypeScript
│ │ └─
video-input.test.ts
TypeScript
│ ├─
package.json
JSON
│ └─
tsconfig.json
JSON
├─
▾
references
│ ├─
cli.md
Markdown
│ └─
workflows.md
Markdown
├─
▾
scripts
│ ├─
chat.sh
Shell
│ ├─
image.sh
Shell
│ ├─
models.sh
Shell
│ ├─
prompt.sh
Shell
│ ├─
resolve_nano_gpt_cli.sh
Shell
│ └─
video.sh
Shell
├─
package-lock.json
JSON
├─
package.json
JSON
├─
README.md
Markdown
└─
SKILL.md
Markdown
Dependencies 2 items
| Package | Version | Source | Known Vulns | Notes |
|---|---|---|---|---|
commander | ^14.0.1 | npm | No | Caret range - no known vulnerabilities |
typescript | ^5.8.2 | npm | No | Dev dependency, only used at build time |
Security Positives
✓ SKILL.md comprehensively documents all capabilities and guardrails
✓ API key is only transmitted to the declared nano-gpt.com endpoint
✓ File access is limited to user-specified paths via --image/--video/--output flags
✓ Guardrails explicitly warn against uploading sensitive files
✓ No credential harvesting or exfiltration behavior
✓ No arbitrary code execution or shell injection vectors
✓ Config redaction implemented for apiKey display (config.ts redactConfig)
✓ Test credentials are clearly test fixtures with fake values
✓ External URLs in test files use example.com/example.org domains
✓ Scripts use set -euo pipefail for safe shell execution
✓ Buffer.from(b64_json, 'base64') is legitimate image decoding, not obfuscation