Skip to content

MPowerUp — Tool Evaluation Tracker

Tracks tools under research. Status lifecycle: newtriedevaluatedadopted / rejected. Trust axes: Private / Autonomous / Secure (yes / partial / no). Architecture axes: Zone (Arch) = by-design network zone, Zone (Ops) = how it behaves in practice today (1 = P2P, 2 = client–relay, 3 = federated, 4 = centralized / walled garden; a tool may span zones). Operator = who runs the infrastructure (self / third-party / mixed). Policy = strength/shape of policy- vs cryptographically-enforced privacy (freeform). Protocols = underlying protocol(s). Score is 0–10 (your overall fit rating). Deployment substrate (cloud vs self-host) is a separate axis from these topology zones — see the companion cloud-providers.md sovereignty matrix.

Tool Category Protocols Zone (Arch) Zone (Ops) Operator Status Private Autonomous Secure Policy Score Notes Updated
Apache AGE Graph (Postgres ext) new partial partial yes Adds graph queries to PostgreSQL; best if already on Postgres. 2026-06-13
Apache Jena / Fuseki RDF triplestore new partial partial yes RDF triples + SPARQL; formal vocabularies (schema.org/FOAF). Likely overkill unless interoperability is a goal. 2026-06-13
ArangoDB Multi-model DB new partial partial yes Graph + document + key-value; good if resources are messy documents. Apache-2.0 / BSL depending on version. 2026-06-13
Berty P2P BLE mesh evaluated partial yes partial 5.5 Best non-Tor offline BLE mesh + permissive (Apache/MIT). Wesh crypto self-declared not-hardened; study the transport pattern, not the crypto. 2026-06-13
Blazegraph RDF triplestore rejected partial partial yes RDF + SPARQL, reasoning-capable, BUT effectively unmaintained since ~2020 (team moved to Amazon Neptune). Avoid for new work. 2026-06-13
Bluesky Federated social (atproto) AT Protocol (atproto) 3 4 mixed new no partial no weak in practice; portability + credible-exit by DESIGN (DID), not by policy promise Federated by design: atproto splits roles Mastodon bundles - Personal Data Servers (identity/repo), Relays (crawl whole network into one firehose), App Views (index/present); anyone can run any piece. DID identity gives credible-exit + provider portability, stronger than Mastodon where identity is tied to the home instance. BUT in practice nearly everyone runs on Bluesky-the-company PDS/relay/app today, so it behaves like z4 (X-like). Relay crawls the entire network (heavy to run), concentrating power. Separate, ActivityPub-incompatible universe from the fediverse. Decentralization-capable but currently centralized. 2026-06-15
Briar P2P messaging Bramble (over Tor / BLE / WiFi) 1 1 none tried yes yes yes 8.5 Only true offline tier: BLE/WiFi mesh + sneakernet + Mailbox. Cure53-audited. GPLv3; feed-style groups; battery. Strongest offline candidate / interim referral. 2026-06-15
Conversations (XMPP) Federated messaging evaluated no partial partial 4.5 Best budget-Android federated client; OMEMO 1:1 mature, group E2E experimental. Server-bound. 2026-06-13
Cwtch Tor group messaging evaluated yes yes yes 7.0 Best group-metadata model (untrusted discardable group servers); permissive libcwtch (embeddable). Tor-only no offline; heaviest battery; funding-fragile. 2026-06-13
Delta Chat Email-based messaging evaluated partial partial yes 6.5 No-new-infrastructure exemplar; E2E default incl groups; Autocrypt v2 adds FS/PQ; near-zero header metadata. Needs a mailbox; no local mesh. 2026-06-13
did:key Identity adopted yes yes yes 8.5 Current identity (Ed25519). Keep; add Verifiable Credentials (SD-JWT-VC) in Phase 4. 2026-06-13
DuckDuckGo Centralized search HTTPS 4 4 third-party tried partial no yes policy-based no-log promise; receives your IP; no cryptographic guarantee 4.0 Policy-based privacy; centralized; receives your IP. 2026-06-15
Facebook Social (walled garden) closed / proprietary 4 4 third-party (Meta) rejected no no yes surveillance-advertising model; social graph locked in 1.0 Maximal walled garden; closed protocol; social graph locked in. The negative baseline the MPowerUp thesis reacts against - tracked as a contrast anchor, not a candidate. 2026-06-15
GoToSocial Federated social (minimal) ActivityPub 3 3 self evaluated partial partial yes 5.0 Single-binary self-host on Pi-class (Second Boot refurb-server reference); no E2E messaging. 2026-06-15
Iroh P2P transport (QUIC) evaluated partial yes yes 7.5 Best libp2p alternative; Rust core + official Android/iOS FFI. Highest-value transport spike; verify RN binding maturity. 2026-06-13
Jami P2P comms OpenDHT, SIP/TLS 1 1 none evaluated partial yes yes 6.0 Active/mature but IP exposed to peers via DHT + battery-heavy. NOT adopted (corrects prior framing); hybrid verdict adopts no single app. 2026-06-15
Jitsi Video conferencing WebRTC (SFU: Jitsi Videobridge) 1 (1:1) / 4 (multi-party) 1 / 3 (self-hosted bridge) / 4 (meet.jit.si) mixed rejected partial no yes architecture-determined, not policy-based 4.0 Spans zones by call type AND hosting: 1:1 is true browser-to-browser P2P (z1); multi-party routes through the Jitsi Videobridge SFU. On meet.jit.si the bridge is third-party (z4); self-hosting the bridge shifts trust to you (behaves like z3). Not E2E in multi-party by default - the bridge must route streams. Dropped from MPowerUp precisely because it does not sit in one zone. 2026-06-15
Kuzu Embedded graph DB embedded (no network) 1 1 none (embedded, data axis) new yes yes partial 7.0 MIT, embedded, no server; runs inside app. 2026-06-15
libsignal Encryption protocol rejected yes yes yes 4.0 Gold-standard 1:1 but AGPL + group-shape mismatch. Borrow ideas (Double Ratchet, PQXDH), not the code. 2026-06-13
Logseq Knowledge graph (local) local Markdown files 1 1 none (local, data axis) new yes yes partial 8.0 Open-source, local Markdown; fits data-sovereignty. 2026-06-15
Mastodon / ActivityPub Federated social ActivityPub 3 3 mixed rejected partial partial no 3.0 No E2E DMs in the family; public-social only. Wrong tool for private messaging; possible future open/Z3 broadcast layer. 2026-06-15
Matrix / Element Federated messaging rejected no partial yes 4.0 Server mandatory (fails offline/no-server). But mine matrix-sdk-crypto (network-free Olm/Megolm) as a reusable component. 2026-06-13
Memgraph Graph DB (server) new partial partial yes Neo4j-compatible (Cypher), in-memory/fast, real-time. Source-available license — check against open-source-only standard. 2026-06-13
Meshtastic Off-grid LoRa mesh new yes yes yes 7.0 Off-grid bearer over LoRa; BLE bridge from RN. Needs per-user LoRa hardware (Second Boot build). Pilot for off-grid. 2026-06-13
Neo4j Community Graph DB (server) new partial partial yes 6.0 Self-host; GPLv3 Community edition; great visualization. 2026-06-13
Nostr / Damus Client-relay social Nostr 2 2 third-party (relays; self-hostable) evaluated partial partial partial 5.0 No PFS; persistent pubkey + public timestamps leak DM metadata. Value is embeddability (nostr-tools MIT) for a future public layer, not privacy. 2026-06-15
Obsidian Knowledge graph (local) new yes partial partial Graph view, plain Markdown, on-device. NOT open source (free for personal use) — counts against the open-source-only standard. (The BNI vault itself is an Obsidian vault.) 2026-06-13
OnionShare Anonymous file-share evaluated yes yes yes 6.0 Ephemeral anonymous file/secret-drop primitive over Tor; not a persistent messenger. 2026-06-13
OpenMLS (MLS) Group encryption protocol evaluated partial yes yes 7.5 RFC 9420; forward secrecy + post-compromise security over the shared-key model. Permissive. Gate = decentralized delivery + FFI. 2026-06-13
PeerTube Federated video publishing ActivityPub (+ WebTorrent P2P delivery) 3 3 self / mixed new partial partial yes per-instance operator trust; public content (no E2E role) 6.0 ActivityPub video (YouTube analog) with WebTorrent P2P delivery to offload bandwidth; federated/self-hostable. Relevant to MPowerUp only if scope grows to video publishing - out of current comms scope. Public-by-nature; no private-messaging role. 2026-06-15
Proton Centralized E2EE suite OpenPGP, WireGuard 4 4 third-party new partial no yes zero-access E2EE on content (cryptographic guarantee); policy-only on metadata Zero-access E2EE: servers cannot read content even though centralized - a structural guarantee, not a no-log promise (vs DuckDuckGo). Stays z4 because: single operator + account lock-out; sees metadata (recipients, timing, subject for non-Proton mail, IP unless using their VPN/Tor); full E2EE only Proton-to-Proton (drops to standard transport to Gmail etc.); Swiss legal compulsion can target metadata. Proof z4 is not monolithic - content confidentiality and metadata exposure are separate axes. 2026-06-15
Quiet Tor + libp2p team chat evaluated yes yes partial 5.0 Closest prior-art to MPowerUp (libp2p + CRDT over Tor). Self-described not audited; heavy stack. 2026-06-13
Scuttlebutt (SSB) P2P social SSB gossip 1 1 none rejected no yes partial 3.5 Permanent un-deletable logs + open social graph unsafe for vulnerable users; Manyverse winding down. 2026-06-15
Session Onion messaging evaluated partial partial partial 4.5 Best UX/battery but NO forward secrecy today + 2026 Oxen consensus-takeover paper (eprint 2026/773) + token dependency. Do not use as trust anchor. 2026-06-13
Signal Encrypted messaging Signal Protocol (Double Ratchet, X3DH/PQXDH) 4 4 third-party new partial no yes content E2E by crypto; minimal metadata + sealed sender by POLICY; centralized operator 7.0 Best-in-class E2E content (Double Ratchet + post-quantum PQXDH). But centralized single operator (Signal servers mandatory) and phone-number identity links real-world ID; sealed sender + minimal retention reduce metadata by policy, not architecture. Complements libsignal already in tracker (borrow ideas, AGPL) - the app itself is z4. 2026-06-15
SimpleX Chat Secure messaging (no-identifier) evaluated yes partial yes 8.0 No persistent identifiers; best metadata resistance; small-group model fits Circles. AGPLv3 + Haskell hard to embed in RN; no local mesh; battery. Design exemplar / interim referral. 2026-06-13
Threads Social (fediverse edge) ActivityPub (partial, Meta-operated) 3 (partial ActivityPub) / 4 (Meta core) 4 third-party (Meta) new no partial yes Meta data-driven; embrace-extend concern over ActivityPub 3.0 Meta-owned; federates outbound via partial ActivityPub but the core is a walled garden. Straddles federated to walled-garden - classic embrace-extend concern (absorb fediverse reach, then wall it). Contrast/watch point, not a candidate. 2026-06-15
Tor / Arti Anonymity network / transport evaluated yes partial yes 7.5 Add network-identity (IP) axis via Arti client crate (MIT/Apache). Mission-aligned; battery cost on mobile; do not host onion services from Arti yet. 2026-06-13
Tox P2P messaging rejected partial yes no 2.5 Unaudited/alpha crypto + IP exposed to contacts. Reject for vulnerable users. 2026-06-13
Veilid Privacy P2P framework new yes yes yes 4.5 Strong privacy philosophy but demo-grade + Flutter-first (no RN binding). Watch. 2026-06-13
WebRTC Transport adopted partial partial yes 7.0 Current baseline transport (DTLS). Keep; benchmark Iroh against it. Needs TURN/relay for symmetric NAT. 2026-06-13
X / Twitter Social (walled garden) closed / proprietary 4 4 third-party rejected no no yes single-company silo; no federation 2.0 Single-company silo; no federation; the centralized baseline the decentralization thesis reacts against. Contrast anchor, not a candidate. 2026-06-15
Yjs CRDT sync engine adopted yes yes partial 8.0 Current stack; RN-proven per-Circle docs. Keep. Read Cinapse anti-CRDT counter-evidence before doubling down. 2026-06-13