메뉴
Run a periodic IAM access review — lists all users, permissions, activity, and flags issues. Required quarterly for SOC 2.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Deep code scan for AI security issues — prompt injection, PII in prompts, hardcoded keys, unguarded agents.
Run AI governance checks across cloud accounts and code repos — ISO 42001, EU AI Act, NIST AI RMF compliance.
Scan cloud accounts and GitHub repos to discover AI/ML services and build an AI system inventory.
Walk staged changes against the engineering principles checklist and report pass/fail per principle. Run before any non-trivial commit. Catches doc drift, stub functions, single-region defaults, missing framework mappings, and other regressions before they ship.
Generate a public-facing security trust page from scan data. Produces a single deployable index.html that shows compliance framework scores, security policies, infrastructure overview, and data protection posture. Deployable to S3, Vercel, Netlify, or GitHub Pages.
Paste a vendor's domain. Get a security risk assessment in 60 seconds.
| name | review-access |
| description | Run a periodic IAM access review — lists all users, permissions, activity, and flags issues. Required quarterly for SOC 2. |
| user-invocable | true |
Conduct a quarterly IAM access review for SOC 2 compliance (CC6.2, CC6.3).
Read shasta.config.json for python_cmd. Use that for all commands (shown as <PYTHON_CMD>).
Run the access review:
<PYTHON_CMD> -c "
import json
from shasta.config import get_aws_client
from shasta.workflows.access_review import run_access_review, save_access_review
client = get_aws_client()
client.validate_credentials()
print('Running access review...')
report = run_access_review(client)
path = save_access_review(report)
print(json.dumps({
'report_path': str(path),
'total_users': report.total_users,
'users_with_console': report.users_with_console,
'users_with_mfa': report.users_with_mfa,
'users_flagged': report.users_flagged,
'flagged_users': [{
'username': r.username, 'flags': r.flags,
'has_console': r.has_console, 'has_mfa': r.has_mfa,
'groups': r.groups, 'attached_policies': r.attached_policies,
'days_inactive': r.days_inactive,
} for r in report.records if r.flags]
}, indent=2))
"
Present interactively: summary stats, then each flagged user with recommended action (keep/modify/remove). The saved report serves as audit evidence.