Noninvasive Input
Brain2Qwerty decoded sentences, typed prompts, voice transcripts, image notes, video notes, and GUI observations are merged into one auditable intent stream.
Brain-Side Mesh Thesis
Brain2Qwerty proves that noninvasive brain recordings can be decoded into typed sentences. Meta Mesh for the Brain turns that output into structured input for the Solana harness: RPC_URL becomes live chain awareness, TRIBE-style multimodal context frames the research layer, and every market path stays approval-gated.
This is not a claim of sentience or direct mind reading. It is a practical interface between noninvasive decoded text, real-time blockchain telemetry, market data, AI tool calling, and a Solana operator surface.
Meta Mesh Agent Layer
Brain2Qwerty decoded sentences, typed prompts, voice transcripts, image notes, video notes, and GUI observations are merged into one auditable intent stream.
TRIBE v2-style video, audio, and language modeling and DECOMEG typing data anchor the brain side before decoded text enters any Solana workflow.
RPC_URL drives slot, block height, epoch, node health, version, and performance sampling so the agent can reason from live chain state.
Helius DAS resolves token metadata while Birdeye adds price and liquidity evidence before any recommendation is drafted.
xAI function calling, realtime search, and public media URLs can enrich the response through allowlisted Solana tools, but the harness never holds keys, signs transactions, or executes trades by default.
Vulcan command plans support read-only market data, paper TWAP/grid/TA loops, dry runs, and explicitly gated live modes for Phoenix perpetual futures.
Plasma is recorded as sandwich-resistant Solana AMM context with program-id and audit metadata, not as an automatic swap executor.
OPENGATEWAY_KEY adds OpenAI-compatible chat analysis to the backend; the browser only asks for server-side model evidence.
RPC_URL Run
The monitor and mission commands turn a normal Solana RPC endpoint into continuous input for the Meta Mesh. They poll the chain, format the observation, attach media evidence, then feed everything into the approval-gated proposal layer.
export RPC_URL="https://api.mainnet-beta.solana.com"
export HELIUS_API_KEY="..."
export BIRDEYE_API_KEY="..."
export XAI_API_KEY="..."
export OPENGATEWAY_KEY="..."
export OPENGATEWAY_MODEL="mimo-v2.5-pro"
export GOBOT_REPO_PATH="/Users/8bit/clawdbot-go"
python -m solana_robot_harness.cli parse-brain2qwerty \
--input ./predictions_test.csv
python -m solana_robot_harness.cli preflight
python -m solana_robot_harness.cli preflight \
--live-opengateway-check
python -m solana_robot_harness.cli run-monitor \
--token <MINT> \
--intent "watch liquidity and chain health" \
--brain-file ./predictions_test.csv \
--iterations 3 \
--interval 5
python -m solana_robot_harness.cli mission \
--token <MINT> \
--brain-file ./predictions_test.csv \
--audio-file ./voice-command.wav \
--transcribe-audio \
--image-file ./chart.png \
--video-file ./screen.mp4 \
--gui-file ./terminal.txt \
--media-url "https://example.com/public-chart.png" \
--use-xai \
--use-opengateway \
--upload-local-media \
--phoenix-symbol SOL \
--phoenix-notional-usdc 25 \
--phoenix-mode paper \
--include-plasma \
--ledger ./runs/metaclawd-missions.jsonl
python -m solana_robot_harness.cli approval-queue \
--ledger ./runs/metaclawd-missions.jsonl
python -m solana_robot_harness.cli sentinel \
--profile ./sentinel-profile.jsonResearch Mesh
A tri-modal foundation model points toward in-silico neuroscience: video, audio, and language features can be related to predicted fMRI responses across subjects and tasks.
The SpanishBCBL release grounds the Brain2Qwerty side: healthy adults read, waited, then typed memorized Spanish sentences while noninvasive signals and keystrokes were aligned.
The deployed API keeps provider keys on the server, merges decoded text with Solana observations, and returns an approval-gated operator response for any compatible frontend or robot.
/api/mesh/analyze/api/mesh/statusPhoenix Perps + Plasma
Meta Mesh mission runs can attach Vulcan strategy commands in paper mode, so agents can test against live prices without wallet or chain activity.
confirm-each and auto-execute modes require explicit flags, and auto-execute also requires a notional guardrail before the command can run.
The Plasma profile is read-only Solana venue evidence for sandwich-resistant AMM routing decisions; the harness does not build Plasma swaps.
python -m solana_robot_harness.cli phoenix-market \
--kind ticker \
--symbol SOL
python -m solana_robot_harness.cli phoenix-perps-plan \
--strategy twap \
--symbol SOL \
--side buy \
--notional-usdc 50 \
--slices 5 \
--interval-seconds 60 \
--mode paper
python -m solana_robot_harness.cli phoenix-perps-plan \
--strategy grid \
--symbol SOL \
--center-on-mark \
--width-pct 1.5 \
--levels-per-side 3 \
--tokens-per-level 0.1 \
--mode dry-run
python -m solana_robot_harness.cli plasma-profile \
--inspect-rpcBrain2Qwerty Evidence
Brain2Qwerty v2 moves the signal from keystroke reconstruction toward complete sentence generation from noninvasive MEG. That makes it the right substrate for a Solana agent that listens to decoded language instead of raw keyboard input.
Front-End Demo
The simulator mirrors the backend mission shape: decoded text, voice or GUI notes, public media URLs, xAI or GitLawb OpenGateway model evidence, automation context, one deterministic optimal_response, a rendered evidence matrix, and operator queues that stay approval-gated.
Decoded Sentences
These examples show the underlying Brain2Qwerty behavior that Meta Mesh treats as decoded text input before any Solana reasoning begins.
Meta Mesh Studies
RPC_URL chain state.
Safety Contract
Meta Mesh for the Brain is designed as a Solana-first research and operator interface. It can monitor, analyze, and propose, but it does not custody keys, sign transactions, or claim autonomous financial authority.