mobile-security

name	mobile-security
description	Implement mobile application security for iOS and Android. Outputs secure storage, certificate pinning, code obfuscation, biometric auth, and API security patterns.
argument-hint	["platform","data sensitivity","authentication method","compliance requirements"]
allowed-tools	Read, Write

Mobile Security

Mobile apps face unique threats: rooted/jailbroken devices, reverse engineering, insecure local storage, and network interception. Defence requires secure storage, certificate pinning, code hardening, and proper authentication — not just HTTPS.

Secure Storage

// iOS — Keychain for secrets
import Security

class KeychainStorage {
    func save(key: String, data: Data) -> Bool {
        let query: [String: Any] = [
            kSecClass as String:            kSecClassGenericPassword,
            kSecAttrAccount as String:      key,
            kSecValueData as String:        data,
            kSecAttrAccessible as String:   kSecAttrAccessibleWhenUnlockedThisDeviceOnly,
        ]
        SecItemDelete(query as CFDictionary)  // Delete existing
        return SecItemAdd(query as CFDictionary, nil) == errSecSuccess
    }
    
    func load(key: String) -> Data? {
        let query: [String: Any] = [
            kSecClass as String:       kSecClassGenericPassword,
            kSecAttrAccount as String: key,
            kSecReturnData as String:  true,
            kSecMatchLimit as String:  kSecMatchLimitOne,
        ]
        var result: AnyObject?
        SecItemCopyMatching(query as CFDictionary, &result)
        return result as? Data
    }
}

// NEVER use UserDefaults for sensitive data
// UserDefaults.standard.set(authToken, forKey: "token") — BAD

// Android — EncryptedSharedPreferences
import androidx.security.crypto.EncryptedSharedPreferences
import androidx.security.crypto.MasterKeys

val masterKey = MasterKeys.getOrCreate(MasterKeys.AES256_GCM_SPEC)

val securePrefs = EncryptedSharedPreferences.create(
    "secure_prefs",
    masterKey,
    context,
    EncryptedSharedPreferences.PrefKeyEncryptionScheme.AES256_SIV,
    EncryptedSharedPreferences.PrefValueEncryptionScheme.AES256_GCM
)

// Save token securely
securePrefs.edit().putString("auth_token", token).apply()

// Android Keystore for cryptographic keys
val keyGenerator = KeyGenerator.getInstance(KeyProperties.KEY_ALGORITHM_AES, "AndroidKeyStore")
keyGenerator.init(
    KeyGenParameterSpec.Builder("my_key_alias",
        KeyProperties.PURPOSE_ENCRYPT or KeyProperties.PURPOSE_DECRYPT)
        .setBlockModes(KeyProperties.BLOCK_MODE_GCM)
        .setEncryptionPaddings(KeyProperties.ENCRYPTION_PADDING_NONE)
        .setUserAuthenticationRequired(true)   // Require biometric
        .build()
)

Certificate Pinning

// iOS — URLSession with certificate pinning
import Foundation
import CryptoKit

class PinnedURLSession: NSObject, URLSessionDelegate {
    // SHA-256 hash of the server's public key
    let pinnedPublicKeyHash = "base64encodedSHA256HashHere=="
    
    func urlSession(_ session: URLSession,
                    didReceive challenge: URLAuthenticationChallenge,
                    completionHandler: @escaping (URLSession.AuthChallengeDisposition, URLCredential?) -> Void) {
        
        guard challenge.protectionSpace.authenticationMethod == NSURLAuthenticationMethodServerTrust,
              let serverTrust = challenge.protectionSpace.serverTrust,
              let certificate = SecTrustGetCertificateAtIndex(serverTrust, 0),
              let publicKey = SecCertificateCopyKey(certificate),
              let publicKeyData = SecKeyCopyExternalRepresentation(publicKey, nil) as Data? else {
            completionHandler(.cancelAuthenticationChallenge, nil)
            
        }
        
         serverHash  .hash(data: publicKeyData).base64EncodedString()
        
         serverHash  pinnedPublicKeyHash {
            completionHandler(.useCredential, (trust: serverTrust))
        }  {
            completionHandler(.cancelAuthenticationChallenge, )
        }
    }
}

Biometric Authentication

// iOS — Face ID / Touch ID
import LocalAuthentication

func authenticateWithBiometrics() async -> Bool {
    let context = LAContext()
    var error: NSError?
    
    guard context.canEvaluatePolicy(.deviceOwnerAuthenticationWithBiometrics, error: &error) else {
        return false
    }
    
    do {
        return try await context.evaluatePolicy(
            .deviceOwnerAuthenticationWithBiometrics,
            localizedReason: "Authenticate to access your account"
        )
    } catch {
        return false
    }
}

Anti-Patterns to Avoid

Anti-Pattern	Problem	Fix
Tokens in SharedPreferences/UserDefaults	Readable on rooted device	EncryptedSharedPreferences / Keychain
No certificate pinning	MITM attacks via rogue CA	Pin public key hash; handle pin rotation
Hardcoded API keys in binary	Extractable via reverse engineering	Server-side key management; device attestation
Logging sensitive data	Crashes logs sent to third-party analytics	Strip PII/tokens from all log statements
No jailbreak/root detection	Compromised device bypasses security controls	Detect and warn/block on compromised devices

10 Rules

1. Keychain (iOS) / EncryptedSharedPreferences + Keystore (Android) for all secrets.
2. Never log authentication tokens, PII, or financial data.
3. Certificate pinning for all production API connections — with a rotation plan.
4. Biometric authentication for high-value actions — not just app unlock.
5. Reverse engineering hardening: minification, obfuscation, anti-tamper checks.
6. Detect rooted/jailbroken devices and degrade security-sensitive features.
7. OWASP Mobile Top 10 guides the security test checklist.
8. Static analysis (MobSF, semgrep) in CI catches common mobile security issues.
9. Dynamic analysis (Frida, Objection) as part of penetration testing.
10. Secure communication: TLS 1.2+, certificate pinning, no plaintext fallback.

---

Deep Reference Playbook

The sections below extend this skill into a complete operating playbook so it can run end-to-end inside Claude Code, CoWork, or any agentic tool without further prompting. Pull only the sections you need for a given engagement.

Inputs the skill must collect

Before producing any output, the skill confirms:

1. Objective — the single decision or artifact the user wants out of this session.
2. Context — system, team, customer, product, or domain the work sits inside.
3. Constraints — time, budget, headcount, regulatory, technical, political.
4. Definition of done — what "good" looks like and who signs it off.
5. Audience — who reads or consumes the output (engineer, exec, customer, regulator).
6. Existing artifacts — prior versions, related docs, dashboards, tickets.
7. Risk appetite — how reversible the decision is and how much ambiguity is acceptable.

If any of these are missing, the skill asks targeted clarifying questions before generating output. It never invents constraints the user did not state.

Operating workflow

The canonical workflow for Mobile Security runs in five stages. Each stage has an explicit exit criterion so the skill knows when to advance.

Stage 1 — Frame. Restate the problem in one paragraph. Name the decision, the deadline, the stakeholders, and the success metric. Surface assumptions explicitly so they can be challenged.

Stage 2 — Diagnose. Inventory the current state with concrete evidence: metrics, quotes, screenshots, configs, tickets. Separate facts from interpretations. Identify the two or three root causes that explain most of the gap, not the long tail of symptoms.

Stage 3 — Design. Generate at least two viable options. For each option, capture: what changes, who owns it, what it costs, what it unblocks, what it risks, and how it could fail. Recommend one with a written rationale.

Stage 4 — Execute. Convert the chosen option into a sequenced plan: milestones, owners, dependencies, gating checks, communication cadence, and rollback triggers. Anything that cannot be assigned an owner and a date is not yet a plan.

Stage 5 — Validate. Define how success will be measured, when the measurement happens, and what action follows each possible result. Schedule the retrospective before the work starts, not after.

Outputs the skill produces

Depending on the request, the skill returns one or more of:

• A one-page brief suitable for an executive reader.
• A detailed working document for the delivery team.
• A decision record capturing the choice, the alternatives, and the rationale.
• A risk register with probability, impact, owner, and mitigation.
• A sequenced action plan with named owners and explicit due dates.
• A measurement plan tied to the success metric.
• A communication plan for stakeholders inside and outside the team.

Every artifact uses clear headings, short paragraphs, and tables where comparison helps. No filler. No restating the prompt. No hedging language when a recommendation is warranted.

Decision logic and trade-offs

The skill applies the following heuristics when choices are not obvious:

• Prefer reversible decisions taken quickly over irreversible decisions taken slowly.
• Optimise for the constraint that bites first — usually time, attention, or trust, not money.
• Default to the simplest design that meets the stated definition of done; add complexity only when a specific requirement forces it.
• Make the cost of being wrong visible so the reader can judge whether the recommendation is proportionate.
• Name the people, not the roles, when assigning ownership; ambiguous ownership produces ambiguous outcomes.

Anti-patterns the skill refuses to emit

Anti-pattern	Why it fails	What the skill does instead
Generic best-practice list with no context	Reader cannot act on it	Tailors recommendations to the stated constraints
Recommendation without trade-offs	Hides the cost of being wrong	Names the price paid for the recommendation
Plan with no owners or dates	Cannot be executed or tracked	Assigns a named owner and a date to every action
Metrics theatre	Measures activity, not outcome	Ties every metric back to the user or business outcome
Boil-the-ocean scope	Nothing ships	Cuts scope to the smallest valuable slice
Buried recommendation	Reader misses the point	Leads with the recommendation in the first paragraph

Quality bar

The skill self-checks each output against these gates before returning it:

1. Can a busy executive understand the recommendation from the first 150 words?
2. Is every claim either evidenced, labelled as an assumption, or removed?
3. Does every action have an owner and a date?
4. Are the trade-offs of the recommendation stated honestly?
5. Is there a measurable success criterion?
6. Would the author be comfortable defending this artifact in a review meeting?

If any gate fails, the skill rewrites the section before returning it.

Worked micro-example

Context: Implement mobile application security for iOS and Android. Outputs secure storage, certificate pinning, code obfuscation, biometric auth, and API security patte

Frame: the team needs a defensible recommendation within five working days; the audience is a cross-functional steering group; the cost of delay is higher than the cost of being slightly wrong.

Diagnose: the dominant constraint is decision latency, not analytical depth. Existing data is sufficient for a directional call.

Design: two viable options surfaced. Option A optimises for speed and reversibility. Option B optimises for completeness but slips the deadline by two weeks.

Execute: Option A recommended. Plan sequenced into a two-week sprint with named owners, a mid-point checkpoint, and a clear rollback trigger.

Validate: success measured against a single leading indicator at day 30 and a single lagging indicator at day 90. Retrospective scheduled for day 35.

Cadence and follow-through

A one-shot artifact rarely changes outcomes. The skill recommends a lightweight cadence to keep the work alive:

• Weekly: owner posts a five-line status (done, doing, blocked, risk, ask).
• Fortnightly: steering group reviews leading indicators and unblocks dependencies.
• Monthly: retrospective on what the data is teaching the team; adjust plan accordingly.
• Quarterly: revisit the original objective and decide whether to continue, pivot, or stop.

Closing rules of thumb

1. Lead with the recommendation; supporting analysis follows.
2. Treat every output as a draft that will be challenged; pre-empt the obvious objections.
3. Prefer one strong recommendation over three weak options.
4. When the evidence is thin, say so; do not launder uncertainty as confidence.
5. Optimise for the next decision, not for the perfect document.
6. Make it easy for the reader to disagree with you in a structured way.
7. Ship the artifact; iterate against feedback rather than in private.