hypnguyen1209/offensive-claude / skills/cloud-security
cloud-security
FlaggedUse when attacking AWS/Azure/GCP cloud — IAM/identity privilege escalation, IMDS/metadata SSRF, Entra device-code & PRT theft, GCP impersonation chains, Kubernetes/container escape, IaC/CI-CD federation abuse
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 9.0 KB |
| references/aws-iam-privesc.md | markdown | 10.4 KB |
| references/azure-entra-attacks.md | markdown | 9.2 KB |
| references/gcp-attacks.md | markdown | 8.4 KB |
| references/iac-secrets-ci-cd.md | markdown | 7.7 KB |
| references/imds-metadata-ssrf.md | markdown | 8.7 KB |
| references/kubernetes-container-escape.md | markdown | 8.9 KB |
| scripts/aws_privesc_enum.py | script | 6.6 KB |
| scripts/entra_device_code_phish.py | script | 5.9 KB |
| scripts/gcp_impersonation_mapper.py | script | 6.7 KB |
| scripts/imds_harvester.py | script | 6.9 KB |
| scripts/k8s_can_i_abuse.py | script | 6.9 KB |
| scripts/oidc_trust_auditor.py | script | 7.1 KB |
scripts/gcp_impersonation_mapper.py
View on GitHub#!/usr/bin/env python3
"""
gcp_impersonation_mapper.py - Build the GCP service-account impersonation graph for a project:
who can impersonate whom (roles/iam.serviceAccountTokenCreator, .actAs, key creation), and which
chains reach a high-privilege identity (owner/editor/security/iam admin).
The danger in GCP is rarely a single binding; it is a CHAIN (A -> B -> C -> org admin) where each
edge looks benign. This tool enumerates the edges and reports reachable privileged targets.
USAGE
python3 gcp_impersonation_mapper.py --project TARGET [--out gcp_graph.json]
python3 gcp_impersonation_mapper.py --project TARGET --start you@example.com
WHAT IT CHECKS (read-only)
- Project IAM policy (principals -> roles).
- Per-SA IAM policy: who holds tokenCreator / actAs / serviceAccountKeyAdmin on each SA.
- Builds a directed graph and does BFS from --start (default: active gcloud account) to any SA
that holds a privileged project role.
DEPENDENCIES
pip install google-cloud-iam google-cloud-resource-manager google-auth
OR rely on `gcloud` being installed (the tool falls back to `gcloud ... --format=json`).
Auth: Application Default Credentials or an active gcloud session.
OPSEC
getIamPolicy calls are Admin Activity / Data Access logged. Read-only; performs no impersonation.
"""
import argparse
import collections
import json
import shutil
import subprocess
import sys
PRIV_ROLES = {
"roles/owner", "roles/editor",
"roles/iam.securityAdmin", "roles/iam.serviceAccountAdmin",
"roles/iam.serviceAccountKeyAdmin", "roles/iam.organizationRoleAdmin",
"roles/resourcemanager.organizationAdmin", "roles/resourcemanager.projectIamAdmin",
}
IMPERSONATE_ROLES = {
"roles/iam.serviceAccountTokenCreator", # mint tokens for the target SA
"roles/iam.serviceAccountUser", # actAs (attach to created resources)
"roles/iam.serviceAccountKeyAdmin", # mint long-lived keys for the target SA
"roles/iam.workloadIdentityUser", # federated impersonation
}
def gcloud_json(args):
"""Run a gcloud command returning JSON; exit on failure."""
if not shutil.which("gcloud"):
sys.exit("[!] gcloud not found and SDK fallback unavailable; install gcloud or the SDKs.")
cmd = ["gcloud"] + args + ["--format=json"]
try:
out = subprocess.run(cmd, capture_output=True, text=True, check=True, timeout=120)
except subprocess.CalledProcessError as e:
sys.exit(f"[!] {' '.join(cmd)} failed:\n{e.stderr}")
return json.loads(out.stdout or "[]")
def active_account():
accts = gcloud_json(["auth", "list", "--filter=status:ACTIVE"])
return accts[0]["account"] if accts else None
def list_service_accounts(project):
sas = gcloud_json(["iam", "service-accounts", "list", "--project", project])
return [s["email"] for s in sas]
def project_iam(project):
pol = gcloud_json(["projects", "get-iam-policy", project])
# member -> set(roles)
out = collections.defaultdict(set)
for b in pol.get("bindings", []):
for m in b.get("members", []):
out[m].add(b["role"])
return out
def sa_iam(project, sa_email):
pol = gcloud_json(
["iam", "service-accounts", "get-iam-policy", sa_email, "--project", project]
)
out = collections.defaultdict(set)
for b in pol.get("bindings", []):
for m in b.get("members", []):
out[m].add(b["role"])
return out
def norm(member):
"""Strip member type prefix -> bare identity (user:/serviceAccount:/group:)."""
return member.split(":", 1)[1] if ":" in member else member
def main():
p = argparse.ArgumentParser(description="GCP SA impersonation graph mapper")
p.add_argument("--project", required=True)
p.add_argument("--start", help="starting principal (default: active gcloud account)")
p.add_argument("--out", help="write full graph JSON")
a = p.parse_args()
start = a.start or active_account()
if not start:
sys.exit("[!] no active account; pass --start <principal>")
print(f"[*] Project: {a.project}")
print(f"[*] Start : {start}\n")
proj_roles = project_iam(a.project)
sas = list_service_accounts(a.project)
print(f"[*] {len(sas)} service account(s) found; mapping impersonation edges ...")
# Build directed edges: principal --(can impersonate)--> sa
edges = collections.defaultdict(set) # principal -> set(target_sa)
edge_reason = {} # (principal, sa) -> roles
privileged_sas = set()
for sa in sas:
if PRIV_ROLES & proj_roles.get(f"serviceAccount:{sa}", set()):
privileged_sas.add(sa)
for member, roles in sa_iam(a.project, sa).items():
imp = roles & IMPERSONATE_ROLES
if imp:
principal = norm(member)
edges[principal].add(sa)
edge_reason[(principal, sa)] = sorted(imp)
if privileged_sas:
print(f"[*] Privileged SAs (hold {'/'.join(sorted(PRIV_ROLES))[:40]}...): "
f"{len(privileged_sas)}")
# BFS from start across impersonation edges to any privileged SA
seen = {start}
queue = collections.deque([(start, [start])])
hits = []
while queue:
node, path = queue.popleft()
for tgt in edges.get(node, ()):
if tgt in seen:
continue
seen.add(tgt)
newpath = path + [tgt]
if tgt in privileged_sas or (PRIV_ROLES & proj_roles.get(f"serviceAccount:{tgt}", set())):
hits.append(newpath)
queue.append((tgt, newpath))
if hits:
print(f"\n[+] {len(hits)} impersonation chain(s) reaching a privileged SA:\n")
for path in hits:
print(" " + " -> ".join(path))
for i in range(len(path) - 1):
why = edge_reason.get((path[i], path[i + 1]), [])
print(f" [{path[i]} -> {path[i+1]}] via {', '.join(why)}")
print()
else:
print("\n[-] No impersonation chain from start reaches a privileged SA "
"(check direct project roles separately).")
print("[*] Direct roles on start principal:")
for r in sorted(proj_roles.get(f"user:{start}", set()) | proj_roles.get(f"serviceAccount:{start}", set())):
print(f" {r}")
if a.out:
graph = {
"project": a.project,
"start": start,
"edges": {k: sorted(v) for k, v in edges.items()},
"edge_reasons": {f"{k[0]}|{k[1]}": v for k, v in edge_reason.items()},
"privileged_sas": sorted(privileged_sas),
"chains_to_privileged": hits,
}
with open(a.out, "w") as fh:
json.dump(graph, fh, indent=2)
print(f"\n[*] Graph -> {a.out}")
if __name__ == "__main__":
main()
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics. | Stars: 471k | Updated | Claude CodeCodex | Verified | Install |
| ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first. | Stars: 194k | Updated | Claude CodeCodex | Verified | Install |