Meta Secret Mobile Application

Web site

Meta Secret Web App

MetaSecret Application

Meta Secret is a decentralised password manager that uses advanced encryption and decentralised storage to securely store and manage user data.

Meta Secret does not rely on a master password to grant access to passwords.

Instead, it uses a combination of biometric authentication and secret sharing techniques to provide secure access to user's confidential information.

Meta Secret is designed to operate on your device(s), the data is encrypted to ensure your information remains private and protected, enabling users to access their passwords from any device without compromising security.

Meta Secret also features a decentralised and open-source infrastructure, providing increased security and privacy for users.

Application Design

Application Structure

Password Split

Password Recovery

Activity Diagram

graph TD
    User --> |split password| MSS{MetaSecret}
    MSS --> |split| Hash1
    MSS --> |split| Hash2
    MSS --> |split| Hash3

    User --> |recover password| MSR{MetaSecret}
    MSR --> |read| HH1[Hash1]
    MSR --> |read| HH2[Hash2]
    HH1 --> RecoverAlgo[Meta Secret: Recovery Algorithm]
    HH2 --> RecoverAlgo
    RecoverAlgo --> RP[Recovered Password]

Sequence Diagram

sequenceDiagram
    note over User: Split to 3 shares
    User->>+MetaSecret: Split password

    MetaSecret->>User: show qr1 (of hash1)
    MetaSecret->>User: show qr2 (of hash2)
    MetaSecret->>-User: show qr3 (of hash3)
    User ->> World: save qr codes in different places

    note over User: Recover from 2 shares
    User ->> World: get qr1
    User ->> World: get qr3

    User ->> MetaSecret: recover password
    User -->> MetaSecret: provide qr1
    User -->> MetaSecret: provide qr3
    MetaSecret ->> MetaSecret: recover password
    MetaSecret ->> User: show password

AI-assisted development

This repository defines a phased workflow (plan → implement → test → verify → review → release) with human approval between phases. Canonical rules live in markdown at the repository root; agents and skills mirror the same discipline. Cargo workspace root: meta-secret/ (see PROJECT_CONTEXT.md).

Read first (everyone)

Skills live under .ai/skills/. Subagent prompts live under .ai/agents/. Domain map for web/cli/core/mobile lib: docs/ai-domain-standards.md.

Claude Code

1. Open this repo in Claude Code so it loads .claude/.
2. Use workflow command specs in .ai/commands/ (invoked through Claude entrypoints).

Start a full delivery chain

Run a single phase

3. After each phase, approve the artifact before the next step. Run phases from the main session (no nested subagents).

4. Build/test hints: from meta-secret/, cargo test, cargo clippy; project-wide Docker: docker buildx bake test (see Infrastructure Build below).

Cursor

Cursor does not load Claude slash command folders directly. Use Agent chat; parity: .cursor/commands/README.md.

1. Rules: .ai/rules/ — ai-project-context.mdc pulls the root markdown documents.
2. Invoke a phase: use workflow entrypoints from .ai/commands/.
3. Skills: ask Agent to read SKILL.md under .ai/skills/<name>/ when needed.
4. Pipeline: WORKFLOW.md. FFI changes: coordinate with meta-secret-compose.

Optional: pattern capture

When triggers in WORKFLOW.md apply, run workflow-pattern-capture with skill workflow-pattern-capture (0–2 suggestions or No changes recommended).

Historical note

Older references may point to docs/ai-skills.md. Current entry points are this section and WORKFLOW.md.

Web Application

Meta Secret Web Cli is available on https://meta-secret.github.io

Command Line App

Split secrets:

You can split and restore your secrets by using meta-secret cli app in docker.
Imagine that we want to split top$ecret, then the command will be:

$ mkdir secrets
$ docker run -ti --rm -v "$(pwd)/secrets:/app/secrets" ghcr.io/meta-secret/cli:latest split --secret top$ecret 

It will generate json/qr(jpg) files (shares of your secert) in the secrets directory.

Restore secrets:

When it comes to restore the secret, put json or qr files (shares of your secret) into the secrets directory. Then run in case of qr (if you want restore from json, just pass --from json ):

$ docker run -ti --rm -v "$(pwd)/secrets:/app/secrets" ghcr.io/meta-secret/cli:latest restore --from qr 

Advice for VPS-users

If you don't want to use FileZilla to download QR-codes to see on your computer, you can see them in terminal.

Installation

$ brew install qrencode (MacOS)
$ apt-get install qrencode (Debian/Ubuntu)
$ dnf install qrencode (CentOS/Rocky/Alma)

Showing QR codes in terminal

$ qrencode -t ansiutf8 < meta-secret-1.json

Congrats! Save these codes in secure place!

Below is optional If you would like to extract data from QR's

• Just take a phone to scan QR
• or screenshot the terminal and upload it on this website: webqr.com

Infrastructure Build

The project uses docker buildx bake as its build system. All build targets are defined in docker-bake.hcl.

Build commands

# Build and push all images (meta-server + web)
docker buildx bake --push default

# Build meta-server image
docker buildx bake meta-server-image

# Build web image
docker buildx bake web-image

# Run tests
docker buildx bake test

# Export web-cli dist locally
docker buildx bake web-local

# Build taskomatic-ai
docker buildx bake taskomatic-ai

Taskomatic AI Docker Image

This Docker image contains the aider AI coding assistant configured to use Claude 3.5 Haiku.

Requirements

• Docker installed
• An Anthropic API key

Usage

Run the container with your Anthropic API key as an environment variable:

docker run -it --rm \
  -e ANTHROPIC_API_KEY=your-api-key-here \
  -v $(pwd):/workspace \
  localhost/taskomatic-ai:latest

Additional options

The container is pre-configured with Claude 3.5 Haiku, but you can pass additional arguments to aider:

# Run with a directory mounted and specific files to edit
docker run -it --rm \
  -e ANTHROPIC_API_KEY=your-api-key-here \
  -v $(pwd):/workspace \
  -w /workspace \
  localhost/taskomatic-ai:latest your_file.py another_file.js

Building the image

docker buildx bake taskomatic-ai

Issues and improvements

If you encounter any issues or have suggestions for improvements, please report them to the project maintainers.