Saarthi-AI security_scan

Run runDependencyAudit, runSastScan, and runHoundDogScan and return a concise, prioritized security summary with critical/high findings first. Must use this skill if security scanning is explicitly requested by the user.

install

source · Clone the upstream repo

git clone https://github.com/SAARTHII-AI/Saarthi-AI

Claude Code · Install into ~/.claude/skills/

T=$(mktemp -d) && git clone --depth=1 https://github.com/SAARTHII-AI/Saarthi-AI "$T" && mkdir -p ~/.claude/skills && cp -r "$T/.local/skills/security_scan" ~/.claude/skills/saarthii-ai-saarthi-ai-security-scan && rm -rf "$T"

manifest: .local/skills/security_scan/SKILL.md

source content

Security Scan Skill

Run three independent scanners and summarize results:

```
runDependencyAudit()
```
for package/dependency vulnerabilities
```
runSastScan()
```
for static code findings
```
runHoundDogScan()
```
for privacy/security dataflow findings

Orchestration

For full scans, run scanners in parallel and tolerate per-scanner failures.

const [depResult, sastResult, hounddogResult] = await Promise.allSettled([
  runDependencyAudit(),
  runSastScan(),
  runHoundDogScan(),
]);

const dep = depResult.status === 'fulfilled' ? depResult.value : null;
const sast = sastResult.status === 'fulfilled' ? sastResult.value : null;
const hounddog =
  hounddogResult.status === 'fulfilled' ? hounddogResult.value : null;

Do not fail the whole scan because one scanner errors.

Minimal Response Shape

runDependencyAudit()

metadata.vulnerabilities

{ info, low, moderate, high, critical }

vulnerabilities[]

id

package

severity

fix

source

runSastScan()

results[]

checkId

message

severity

fingerprint

location

runHoundDogScan()

vulnerabilities[]

hash

ruleIds

message

severity

location

privacyViolations

remediation*

Output Expectations

Return concise results instead of dumping full payloads:

Per scanner: status (
```
ok
```
or
```
error
```
) and count by severity.
Top critical/high findings with file path and short message.
A short remediation plan, with risky/breaking changes clearly called out.