android-backend-api-testing
Test the mobile backend API once traffic is visible — broken auth, weak/stateless JWT, IDOR/BOLA, missing authorization, injection, GraphQL introspection, rate-limiting, and OAuth/PKCE flaws. Tier 1. Use after pinning bypass exposes the API; this is where the highest-severity findings live. Aligns to OWASP MASTG MASVS-NETWORK/AUTH. Payloads: payloads/auth-jwt.md, INJECTION-POINTS.md.
AnandJogawade · 已有静态检查记录
安装与使用
运行命令前先阅读来源与依赖说明。复制命令不会自动安装或执行。
npx skills add AnandJogawade/Zazy-Pentest-Plugin-for-Claude --skill android-backend-api-testing技能说明
name: android-backend-api-testing description: "Test the mobile backend API once traffic is visible — broken auth, weak/stateless JWT, IDOR/BOLA, missing authorization, injection, GraphQL introspection, rate-limiting, and OAuth/PKCE flaws. Tier 1. Use after pinning bypass exposes the API; this is where the highest-severity findings live. Aligns to OWASP MASTG MASVS-NETWORK/AUTH. Payloads: payloads/auth-jwt.md, INJECTION-POINTS.md."
Android Backend / API Testing
The mobile backend is a full API and usually the highest-impact target. Once traffic is visible, test it like any web/API target.
Trigger phrases
mobile backend, mobile api, api testing, bola, idor api, jwt mobile, graphql introspection, oauth pkce, broken authorization, mobile endpoint
Method
- Capture the full API surface through the proxy (after
android-network-cleartext-tls+android-ssl-pinning-bypass): enumerate every endpoint, header, token, and parameter (see${CLAUDE_PLUGIN_ROOT}/payloads/INJECTION-POINTS.md). - Broken object-level authz (IDOR/BOLA): swap object IDs across all verbs with ≥2 accounts → chain to
idor. - Auth & session: weak/stateless JWT validation (
alg:none, weak secret, uncheckedaud/iss/exp) →jwt; session fixation; token lifetime/rotation; replay. - Missing/function-level authz (BFLA): reach admin/privileged endpoints as a low-priv user →
access-control. - Injection: SQLi/NoSQL/command/SSRF in API params and JSON keys (
${CLAUDE_PLUGIN_ROOT}/payloads/) →sqli/nosql-injection/command-injection/ssrf. - GraphQL: introspection enabled, field-level authz, batching abuse →
graphql. - OAuth: PKCE presence, redirect-URI allowlist, code/token leakage (correlate with deep-link token theft,
android-deeplink-applink-abuse) →oauth. - Rate-limiting / mass-assignment on sensitive actions →
rate-limiting/mass-assignment.
Modern / rare techniques
- Legacy/weaker mobile endpoints: apps often expose older API versions with fewer checks than the web app — enumerate
/v1,/mobile,/internalpaths. - Client-trusted fields: mobile APIs frequently trust client-sent role/price/flags → mass-assignment / business-logic abuse (
business-logic). - Correlate client-side leaks with server acceptance: a token leaked via deep link/logcat is only a finding if the server accepts it — prove the round trip.
Bypass / evasion
Reuse the app's real tokens/headers/User-Agent from captured traffic so WAF/device-binding checks pass; see ${CLAUDE_PLUGIN_ROOT}/payloads/waf-bypass.md for filtered injection.
Escalation / chaining
Cross-account access, forged tokens, privileged endpoints → account takeover, data breach, privilege escalation (usually the report's top findings). Chains to the full web/API skill set (api-security, idor, jwt, access-control, sqli, graphql, oauth).
Backend & tooling
Preferred Kali path (native Kali / WSL+Kali / Kali-SSH / Kali-MCP / HexStrike). Needs the user's device/emulator for live traffic + Burp/mitmproxy (Repeater/Intruder); route through Burp if a Burp MCP is present and send manual PoCs to Repeater. Windows-local last resort. See zazy-orchestrator.
Verification (background, before reporting)
Re-confirm every candidate independently — second account/session, replay, or OAST (Collaborator/interactsh) for blind cases — and reproduce before marking CONFIRMED. Report true positives only.
Evidence
The request/response proving the flaw (two-session IDOR, forged token accepted, injection output, reachable privileged endpoint), the affected asset, and impact. Remediation: server-side authz on every request; strong token validation/rotation; object- and function-level checks; PKCE + strict redirect allowlists.