guardskills 0.1.0-alpha.2 → 0.1.0-alpha.3

package/README.md

@@ -23,6 +23,14 @@ npx guardskills add https://github.com/vercel-labs/skills --skill find-skills
 3. Computes a risk decision (`SAFE`, `WARNING`, `UNSAFE`, `CRITICAL`, `UNVERIFIABLE`).
 4. Proceeds to `npx skills add ...` only if gate policy allows.
 
+## Security Notice
+
+`guardskills` is an additional security layer on top of `skills.sh`, not a replacement for your own review process.
+
+- `guardskills` does not maintain, control, or guarantee the safety of `skills.sh` or third-party skill repositories.
+- Static analysis reduces risk but cannot detect every threat.
+- A `SAFE` result means "no known high-risk pattern detected," not "guaranteed safe."
+
 ## Current Readiness
 
 - Current stage: **beta-quality**.
@@ -105,6 +113,18 @@ guardskills add owner/repo --skill name \
   --max-total-files 120
 ```
 
+## Local Check (Folder on Disk)
+
+Current CLI release supports GitHub sources directly (`guardskills add <repo>`).  
+If you have a local skill folder, you can run the scanner manually:
+
+```powershell
+$env:SKILL_DIR = "C:/path/to/your/local-skill-folder"
+npx tsx --eval "import fs from 'node:fs'; import path from 'node:path'; import { scanResolvedSkill } from './src/scanner/scan.ts'; import { calculateRiskScore } from './src/scoring/engine.ts'; const root=process.env.SKILL_DIR; if(!root){throw new Error('SKILL_DIR missing')} const files=[]; const walk=(d)=>{for(const e of fs.readdirSync(d,{withFileTypes:true})){const p=path.join(d,e.name); if(e.isDirectory()) walk(p); else files.push({path:path.relative(root,p).replace(/\\\\/g,'/'),content:fs.readFileSync(p,'utf8')});}}; walk(root); const scan=scanResolvedSkill({source:'local',owner:'local',repo:'local',defaultBranch:'local',commitSha:'local',skillName:path.basename(root),skillDir:root,skillFilePath:'SKILL.md',files,unverifiableReasons:[]}); const decision=calculateRiskScore(scan.findings,{hasUnverifiableContent:scan.hasUnverifiableContent,strict:false,trustCredits:0}); console.log(JSON.stringify({level:decision.level,riskScore:decision.riskScore,findings:scan.findings},null,2));"
+```
+
+If the path is wrong, PowerShell returns an error before scan starts.
+
 ## Configuration File
 
 `guardskills` supports repository-local policy and default settings via `guardskills.config.json`.

package/SECURITY.md

@@ -31,4 +31,10 @@ In scope:
 Out of scope:
 
 - vulnerabilities in third-party skill repositories themselves
-- social engineering reports without technical exploit details
+- social engineering reports without technical exploit details
+
+## Limitations
+
+- `guardskills` is an additional security control layer and not a guarantee of safety.
+- We do not maintain or control `skills.sh` or third-party skill repositories.
+- Scanner-based detection can miss novel or heavily obfuscated malware.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "guardskills",
-  "version": "0.1.0-alpha.2",
+  "version": "0.1.0-alpha.3",
   "description": "Security wrapper around skills add",
   "repository": {
     "type": "git",