bubblescan

Offensive security toolkit for Bubble.io applications — 12 modules covering reconnaissance, configuration audit, and data extraction, driven from a single interactive CLI. Includes a working implementation of the Elasticsearch crypto bypass publicly disclosed in April 2025.

Author: @Siin0pe · License: MIT

Features

Reconnaissance (read-only, passive)

• fingerprint — detect Bubble.io, extract appname, session tokens, API keys, CDN/infra
• plugins — enumerate Bubble plugins (first-party + marketplace), optionally enrich from the Bubble plugin store
• datatypes — list every custom data type and its fields from static.js + /init/data
• pages — enumerate Bubble pages via wordlist (live + /version-test/)
• elements — rebuild the UI element tree from dynamic.js
• secrets — scan HTML + bundles for tokens, API keys, URL secrets; verify exposed Google keys

Audit (active, read-only probing)

• config-audit — security headers, public-editor check, live vs /version-test/ diff
• plugin-audit — flag deprecated / leak-prone plugins, detect third-party data-leak hosts
• api-probe — map the Data API and Workflow API surface (meta, obj, wf, swagger)
• files — enumerate S3/CDN URLs, test anon access, probe /fileupload, optional type-fuzz

Exploit

• es-audit — Elasticsearch crypto bypass: probe, analyze, dump, forge, encrypt/decrypt
• workflows — audit workflow API, detect anon-reachable and temp-password-leaking workflows

Reporting

• One Context collects findings across every module
• Export as Markdown, HTML, or JSON from a single report command
• Optional --checkpoint snapshots after each step

Install

Requires Python 3.11+.

pipx install bubblescan

Verify:

bubblescan --version
bubblescan modules

Reports and dumps land under ./out/ by default, so run bubblescan from the directory you want the artefacts written to.

Usage

Three paths, from simplest to most complete.

1. Quick fingerprint

Single module, one-shot, no state to manage. Good for confirming a target is a Bubble.io app and grabbing the appname, session tokens, keys, and infra:

bubblescan run fingerprint https://app.example.com

2. Full audit + report

Run every module (recon → audit → exploit) in the right order and export a structured report:

bubblescan flow full https://app.example.com --export out/report.html

.md, .html, and .json are all supported — the extension picks the format. Add --open to pop the report in your browser when the flow finishes.

Shorter variants:

bubblescan flow recon   https://app.example.com    # passive only
bubblescan flow audit   https://app.example.com    # + active probing
bubblescan flow crypto  https://app.example.com    # ES bypass end-to-end
bubblescan report       https://app.example.com out/report.html   # alias for `flow full --export`

3. Interactive shell

For iterative work — pick modules, inspect findings, export at the end:

bubblescan                                  # launches the REPL

bubblescan ❯ target https://app.example.com
bubblescan ❯ session load session.json          # optional, authenticated session
bubblescan ❯ modules                             # list modules by phase
bubblescan ❯ help es-audit                       # module-specific help
bubblescan ❯ flow recon                          # chain modules
bubblescan ❯ run es-audit analyze --field-leak   # single module with flags
bubblescan ❯ findings                            # review what was captured
bubblescan ❯ report out/session.html

Tab-completion works on commands, modules, and targets. History persists to ~/.bubblescan_history.

Flow presets

Every preset accepts --export <path>, --open, and --checkpoint.

Environment

Documentation

• docs/cli.md — shell command reference
• docs/modules.md — module-by-module reference
• docs/workflows.md — recipes and end-to-end sessions
• docs/architecture.md — code map for contributors
• docs/crypto.md — Elasticsearch crypto protocol

Elasticsearch crypto bypass

Short version: every Bubble SPA encrypts its Elasticsearch requests into a three-part envelope {x, y, z} before sending them, but the entire derivation hinges on a value every client receives in plaintext — the appname slug. The scheme was reverse-engineered and published in April 2025 by Lucca & Pedro (demon-i386/pop_n_bubble, GBHackers coverage).

• Cipher: AES-256-CBC + PKCS7
• KDF: PBKDF2-HMAC-MD5 with 7 iterations, appname as salt
• Constant IV seeds po9 / fl1, identical across every Bubble app
• No authentication on the endpoint itself — Bubble has not issued a patch

bubblescan re-implements the primitives from scratch (no upstream code) and exposes them through es-audit: probe, analyze, dumpone, dumpall, query, encrypt, decrypt. See docs/crypto.md for the full protocol spec and docs/modules.md#es-audit for the subcommand reference.

Contributing

Ideas, bug reports, and new modules are very welcome.

• Bug reports / feature requests: open an issue with the (anonymised) target context, the command you ran, and the output.
• New modules: follow the short guide in docs/architecture.md. Anything dropped into bubblescan/modules/ is auto-discovered.
• Pull requests: keep them focused, match existing patterns, and add a one-line entry to the relevant docs/*.md when user-visible.

The project is intentionally small — reading three or four modules is enough to get the conventions.

Disclaimer & authorized use

bubblescan is an offensive security research tool. Running it implies acceptance of the terms below.

• Authorized testing only. Use bubblescan only against systems you own or that you have prior written authorization to test (formal engagement, bug-bounty scope, CTF, training lab).
• Unauthorized use is prohibited and is the sole responsibility of the end user. The author accepts no liability for it.
• Public disclosure. The Elasticsearch bypass targets a Bubble.io flaw that was publicly disclosed in April 2025 by Lucca & Pedro; this project re-implements the primitives from the public specification independently.
• No affiliation. bubblescan is not affiliated with, endorsed by, or sponsored by Bubble Group, Inc.
• No warranty. Software provided "as is" per the MIT LICENSE.

Responsible disclosure

• Vulnerability in a Bubble.io app you discover with this tool: report it privately to the application owner with a reasonable fix window (typically 90 days) before any public disclosure.

Credits

• Tool design and implementation: @Siin0pe.
• Cryptographic scheme research: Pablo and Lucca, published April 2025 via demon-i386/pop_n_bubble with coverage from GBHackers, Cyberpress, SecurityOnline, and TechNADU. bubblescan re-implements the primitives independently and wraps them in eleven additional modules covering the rest of the Bubble.io attack surface.