**Open-source counter-surveillance for the networks you control.** Map your exposure. Verify your route. Gate sensitive actions. [![npm](https://img.shields.io/npm/v/antiflock.svg?logo=npm&color=cb3837)](https://www.npmjs.com/package/antiflock) [![PyPI](https://img.shields.io/pypi/v/antiflock.svg?logo=pypi&logoColor=white&color=3775a9)](https://pypi.org/project/antiflock/) [![CI](https://github.com/AetherAI3/AntiFlock/actions/workflows/ci.yml/badge.svg)](https://github.com/AetherAI3/AntiFlock/actions/workflows/ci.yml) [![License: Apache 2.0](https://img.shields.io/badge/License-Apache_2.0-2dd4bf.svg)](LICENSE) [![Status: pre-alpha](https://img.shields.io/badge/Status-pre--alpha-e0a84e.svg)](docs/release-status.md) [![Go 1.26](https://img.shields.io/badge/Go-1.26-00ADD8?logo=go&logoColor=white)](go.mod) [![Node 24](https://img.shields.io/badge/Node-24-339933?logo=node.js&logoColor=white)](package.json) **[Run the demo](#run-the-demo)** · **[Why it exists](#why-antifl0ck-exists)** · **[Architecture](#architecture)** · **[Contribute](#contributing)** · **[Release status](docs/release-status.md)** _{Pre-alpha · The local simulation and dashboard work today. Host-level enforcement remains under active development.}

---

Animated demo: joining an untrusted network holds a sensitive action; after the route is verified the action is allowed and the audit is signed

When the trusted path disappears, AntiFl0ck **holds** the action — records why, verifies the route recovered, and only then lets it through. Every step lands in a signed local audit log. **For you if** you run a homelab, or a Tailscale / Headscale / WireGuard mesh, and you want your devices, routes, and sensitive actions under one operator-defined policy with a signed audit trail. **What it is not:** AntiFl0ck does not detect or locate ALPR cameras, scan for nearby hardware, or interfere with anyone's surveillance infrastructure. It is a defensive layer for the networks *you* already control. **⭐ Star AntiFl0ck** to follow the public build and help more privacy engineers find it. ## Why AntiFl0ck exists Your phone, laptop, accounts, VPN, Wi-Fi networks, and cloud services leave related metadata trails that move together — like a **flock**. Each product shows you one fragment. AntiFl0ck is an open attempt to reveal the whole path, label what is actually known, and let the operator decide what happens next. The debate around Flock Safety and large-scale license-plate-reader (ALPR) networks made one problem impossible to ignore: surveillance infrastructure grows faster than ordinary people's ability to understand their own exposure. AntiFl0ck explores what the *defensive* side should look like. > AntiFl0ck is an independent open-source project. It is not affiliated with or > endorsed by Flock Safety, and it does not interfere with third-party > surveillance infrastructure. ## What it does | | | | --- | --- | | **See** | Map your devices, routes, mesh peers, and the metadata others could observe. | | **Understand** | Label every finding: detected, verified, reported, inferred, suspected, or unknown. | | **Decide** | Deterministic, operator-defined policy allows, holds, or blocks sensitive actions. | | **Prove** | Signed decisions, plain-language explanations, expirations, and rollback results — kept on your machine. |

## Run the demo Fully simulated coffee-shop scenario. Needs **Docker**, **git**, and **Node 20+**; no VPN account or real data. ### npm or pip (fastest — no clone required) ```bash npx antiflock init # clones the repo + generates local config npx antiflock dev # builds and starts Core, the simulator, and the dashboard npx antiflock lab # runs the scripted coffee-shop simulation ``` ```bash pipx install antiflock antiflock init antiflock dev antiflock lab ``` `--help` lists every command (`build`, `down`, `clean`, `--dir`, `--ref`) on either one. [npm package →](https://www.npmjs.com/package/antiflock) · [PyPI package →](https://pypi.org/project/antiflock/) ### From a clone ```bash make dev ``` ```bash make lab ``` On Windows without `make`: `npm run dev` / `npm run lab`. ### Either way Open — username `operator`, token from `.antiflock/dev.env`. Full guide: [operator runbook](docs/operator-runbook.md). ## Working today vs. open The local, simulated slice is complete and sits behind a 10-gate release check (`make verify`). The real-world pieces are separate, later milestones — they are not claimed until proven. **Working today** - Coffee-shop simulation end-to-end, with five signed audit events - Signed, hash-chained event and audit log (SQLite) - Deterministic policy engine, findings, and signed expiring plans - Secure-action gate + TypeScript SDK - Endpoint enrollment bootstrap and mTLS device identity - Linux route/interface observation; opt-in socket-table flow metadata (no packets or payloads) - Read-only live mesh probes — Tailscale CLI and Headscale BYOK — via the durable agent queue - Third-Eye dashboard - Proposal-only Nano watchdog with audited program admission **Open engineering work** - Production enforcement and real packet-path integration - Mobile (Android) enforcement beyond the reference state machine - Provider lifecycle automation (key rotation, revocation) and broader platform collectors - Independent security and privacy review Details and hard boundaries: [release status](docs/release-status.md) · [OPEN decisions and release gates](docs/open-questions.md) · [Explore open work →](https://github.com/AetherAI3/AntiFlock/issues) ## Architecture Core is the brain — identity, events, policy, explanations — but it never sits in your traffic path. Each device keeps a local copy of the rules, so protection keeps working if Core goes offline. ```mermaid flowchart LR A[Devices and providers] -->|Signed observations| B[AntiFl0ck Core] B --> C[Evidence and audit] B --> D[Third-Eye dashboard] B --> E[Policy engine] E -->|Signed expiring plan| F[Local agent] F --> G{Verify result} G -->|Success| H[Allow] G -->|Uncertain| I[Hold or roll back] ``` Deep dives: [architecture](docs/architecture.md) · [threat model](docs/threat-model.md) · [evidence model](docs/evidence-model.md) · [enrolled-agent setup](docs/agent-watchdog-loop.md) · [Nano watchdog](docs/nano-watchdog.md) · [all docs](docs/README.md) ## Contributing Help build the defensive layer surveillance technology never gave ordinary people. Five lanes: - **Observers** — Linux/macOS/Windows/Android collectors, provider adapters - **Visualizers** — topology, exposure maps, audit timelines (React) - **Policy builders** — deterministic rules, secure-action integrations - **Researchers** — metadata models, ALPR research, threat modeling - **Hardening engineers** — signing, storage, privilege separation, testing Start with [CONTRIBUTING.md](CONTRIBUTING.md) and the [good first issues](https://github.com/AetherAI3/AntiFlock/issues?q=is%3Aissue+is%3Aopen+label%3A%22good+first+issue%22). Security and privacy changes begin with a short design note (ADR); how decisions get made is in [GOVERNANCE.md](GOVERNANCE.md). ## Safety, security, and license AntiFl0ck is pre-release defensive software. It does not provide anonymity, prove that surveillance is occurring, or replace a VPN. An AI may *explain* a finding, but it never makes the allow-or-block decision. Report vulnerabilities privately via [SECURITY.md](SECURITY.md), never a public issue. **License:** [Apache-2.0](LICENSE). The AntiFl0ck name and eagle mark are project marks — see [TRADEMARKS.md](TRADEMARKS.md). *AntiFl0ck* is the public name; an earlier working title clashed with another project, so internal identifiers are still being [migrated](docs/REBRAND.md). ---

**Evidence over alarm. Operator over platform.**