Validator / Infra Team — Action Items from Security Audit
Date: 2026-03-30 From: Dev Team Context: We've addressed all code-level findings from the security audit in our repo. The items below require infra team action — either infrastructure changes, access we don't have, or coordination with external services.
1. Deploy Status API Fixes to Production (from your DEV_TEAM_UPDATE.md)
Your endpoint sanitization and real block time fixes are committed but not yet deployed. Per your doc:
This rebuilds the network-status container with sanitized endpoints and real metrics on Sentry-1.
2. Finding #3 (High): Weak Decentralization Signal
What we did: Explorer already displays validator list with voting power and commission.
What we need from you:
Publish a seed node list (node IDs + addresses) that external validators can use to bootstrap
Confirm the active validator set size and current validator count so we can display it accurately
Provide sentry topology guidance — minimum recommended peers, geographic distribution recommendations
Clarify: should we label Makalu as "testnet" explicitly in the explorer UI and status API? The audit flagged that we're presenting testnet posture as mainnet.
3. Finding #4 (High): Binary Provenance
What we need from you:
Publish reproducible build instructions: exact repo URL, commit hash, Go compiler version, and build command for lithod
Publish signed release manifests with SHA256 digests for all official binaries
Document the exact Evmos upstream fork point and patch delta (which Evmos commit was forked, what was changed)
Host deterministic binaries at a single canonical release location (GitHub Releases on KaJLabs/Lithosphere, or a dedicated downloads page)
Generate and publish an SBOM (Software Bill of Materials) — this is planned for Phase 2/3 of the infrastructure roadmap
4. Finding #9 (Medium): NLB Environment Naming
The NLB is named litho-mainnet-rpc-* but serves Makalu testnet. Your doc says this requires downtime to rename.
Request: Schedule this for the next maintenance window. No rush — just don't forget it.
5. Finding #10 (Medium): Rate Limits and Anti-Spam
We've documented what we know in docs/network/chain-parameters.md. But we need authoritative numbers from you:
Minimum gas price enforced by validators (exact value in ulitho)
Mempool configuration: max mempool size, tx queue limits
Nginx/Cloudflare rate limits on rpc.litho.ai and api.litho.ai (requests per second per IP)
Pruning configuration: what's the pruning strategy on sentry nodes? What's the recommendation for indexers that need archive data?
WebSocket connection limits on the EVM WS endpoint
6. gRPC TLS Proxy (from your DEV_TEAM_UPDATE.md)
Your endpoint cleanup changed the default gRPC to grpc.litho.ai:9090. We've used this in our network-parameters.json. Is TLS actually configured on this endpoint now, or is it still direct/plaintext? If plaintext, we should note that in our docs.
Same question for EVM WebSocket — is wss:// available via Nginx, or still ws://54.163.248.63:8546 direct?
7. Open GitHub Issues on KaJLabs/Lithosphere
The audit flagged issue #3 ("[TESTNET] Deployment Failure") — it's now closed. But two issues remain open:
#5: "Fix grammatical error in 'What is Lithosphere?'" (opened 2026-03-24)
Request: Triage these — close, assign, or comment with status. Open unattended issues on a public repo hurt credibility per the audit.
8. Network Parameters JSON
We've published a machine-readable docs/network/network-parameters.json with canonical chain config for wallets and operators. Please review it and confirm all values are correct — especially:
networkType: We set "testnet" — confirm this is correct for Makalu
apis.grpc address: grpc.litho.ai:9090
apis.evmJsonRpc: We pointed at https://rpc.litho.ai — is this the correct public EVM JSON-RPC endpoint?
9. Directory Rename: Makulu → Makalu (CRITICAL for next deploy)
We've renamed the monorepo directory from Makulu/ to Makalu/ to match the network name. All CI/CD pipelines, deploy scripts, and env files now reference /opt/lithosphere/Makalu.
Before the next deploy, you must rename the directory on the production server:
If the deploy runs before this rename, it will create a new /opt/lithosphere/Makalu directory alongside the old one, and the existing containers under Makulu will keep running on stale code.