NEW ANALYSIS
Turn a package into one reverse-engineering and Firebase packet.
Static analysis only. The artifact is never executed, and external checks stay off unless you enable them.
AUTHORIZED ARTIFACTS ONLY
NEW ANALYSIS
Static analysis only. The artifact is never executed, and external checks stay off unless you enable them.
AUTOMATED PIPELINE
WORKSPACE
AWS S3 + DIGITALOCEAN SPACES
FIREBASE + GOOGLE CLOUD
Authenticate one company-owned service account and enumerate only bounded, read-only metadata. The credential is consumed from memory and is never saved in the job.
SAFETY BOUNDARY
ADMINISTRATION
Create isolated teams, rotate scoped access keys, and disable access without deleting reports.
ACCESS KEY
ACCESS BOUNDARIES
AUDIT TRAIL
OPERATOR PLAYBOOK
Upload only an application or archive you own or are authorized to assess. HarborData stores it in a private per-job directory with mode 0600.
Before extraction, the pipeline blocks absolute paths, parent traversal, oversized entries, excessive entry counts, and expansion bombs.
Androguard maps package metadata, permissions, exported Android components, SDK targets and certificate fingerprints. APKRecon extracts high-signal configuration and credential candidates without showing raw values.
JADX creates navigable Java/Kotlin-like source and resources. Split packages are unpacked and handled as one job. Native libraries are inventoried separately.
Firebase project IDs, service-account identities, S3 buckets, Spaces endpoints, domains and URLs are correlated into the report. Discovered credentials stay fingerprint-only unless the separate authorized posture-audit option is enabled.
Network checks are opt-in, rate-limited and read-only. Use the Firebase SA screen for a separately supplied credential, or explicitly enable the discovered-SA option for an owned artifact; writes, deletes and object-body downloads remain blocked.
FIREBASE ASSESSMENT
Live-safe automation is capped, rate-limited, and read-only. Authenticated checks require a separate credential option, server policy and ownership confirmation. Manual-only or destructive checks are never automated by HarborData: no account creation or deletion, password-reset emails, database writes, uploads, callable function invocation, or exploit payloads.
Correlate project IDs, app IDs, API hosts, buckets, database URLs, functions, hosting, and client configuration found in the artifact or public metadata.
Review enabled providers, anonymous access, user-enumeration behavior, password policy, MFA, and tenant settings in an authorized console. Account actions and password-reset flows are out of scope.
Use bounded unauthenticated GET requests against discovered Realtime Database roots. Rules validation that requires writes, payloads, or credentials is manual only.
Perform capped public read and metadata checks only. Collection guessing, document writes, deletes, and rules testing with exploit payloads remain manual only.
Check discovered bucket and object exposure with bounded metadata or HEAD requests. Object uploads, deletes, overwrites, and body downloads are not automated.
Inspect client defaults and make bounded public read checks where applicable. Publishing, activating, or changing templates is manual and destructive.
Discovered Functions and Hosting endpoints are inventoried only. Function invocation is never automated—not with GET, HEAD, POST, callable requests, fuzzing, or exploit payloads.
Review authorized domains and Firebase Dynamic Links configuration in the owned project console. Redirect or domain-takeover validation requires a separately controlled manual plan.
Artifact scans fingerprint discovered key material without using it. The separate Firebase SA audit can authenticate a user-supplied owned credential once to enumerate bounded metadata and test permissions without exercising them.
Review sender IDs, token handling, and server-key hygiene without sending messages or validating credentials. Delivery and authorization tests require a controlled manual procedure.
Checklist reference: Icex0/firebase-pentest-checklist. Use it only within your documented authorization and the safety boundaries above.
Run the static packet first. Triage findings. Enable network checks only for the specific owned environment that needs validation. Rotate every confirmed embedded credential, then rebuild and rescan.