The full upstream README, mirrored here for reference. Install config, tool schemas, adoption signals, and an original overview live on the Hedera MCP Platform listing page.
The intelligence layer for AI agents on Hedera.
23 tools. 7 modules. Pay per call in HBAR. No registration.
HederaToolbox is a production Model Context Protocol server. It gives AI agents structured, metered access to the full Hedera ecosystem — HCS topics, tokens, identity, smart contracts, governance, compliance, and Fixatum DID registration.
Built for agents that need to reason about Hedera, not just interact with it.
Claude Desktop — one business objective, no steps specified, agent decides everything:
Terminal agent — autonomous compliance workflow running headless:
Five production-ready agents in examples/. Clone the repo, fund once, run. Zero dependencies beyond Node.js 18+.
One-time setup for all agents:
Windows users: set env vars before running:
set HEDERA_ACCOUNT_ID=0.0.YOUR_ID && set TOKEN_ID=0.0.731861 && node examples/whale-alert-agent.mjsOr edit the config constants directly at the top of each file.
examples/agent-reputation-monitor.mjs
Watches a list of Fixatum-registered AI agents on a schedule. Detects grade changes and significant score drops — then writes a tamper-proof record to your HCS topic when a change is observed.
Built for platforms that accept agents as counterparties, DAOs monitoring delegate agents, and operators tracking a deployed agent fleet. Polling is free (fixatum_score costs nothing) — you only pay 0.1 HBAR when an alert actually fires.
Create a watched-agents.json file next to the script:
Then run:
First run establishes baselines only — no alerts fire on startup. State is saved to
agents-state.jsonand survives restarts.
| Config | Default | Description |
|---|---|---|
HCS_TOPIC_ID | required | Your HCS topic for alert records |
CHECK_INTERVAL_H | 6 | Hours between checks |
DROP_THRESHOLD | 10 | Point drop that triggers alert regardless of grade |
WATCH_FILE | watched-agents.json | Path to your watch list |
Cost: Free to poll · 0.1 ℏ per alert written to HCS
Agents without a Fixatum DID cannot be monitored. Get one at fixatum.com/register.
examples/compliance-kya-onboarding-agent.mjs
Screens any Hedera account before doing business with them. Runs identity resolution, on-chain risk screening, and optional KYC verification — then pulls a live Fixatum KYA trust score (0–100) for free as part of the same run. The full result is written as a single tamper-proof record to your own HCS topic.
Returns APPROVED, REJECTED, PENDING_REVIEW, or PENDING_KYC.
Designed for token issuers, DEXes, DAOs, and treasury managers who want auditable counterparty screening with a reputation layer, not just a binary pass/fail.
HCS_TOPIC_IDis required — records write to your own topic, not a shared one. If you have a Fixatum DID, youragent_topic_idworks here. Get one at fixatum.com/register.
| Config | Default | Description |
|---|---|---|
SUBJECT | 0.0.7925398 | Account to screen |
HCS_TOPIC_ID | required | Your HCS topic for compliance records |
KYC_TOKEN_ID | none | Optional: your token ID for KYC check |
SKIP_VERIFY | false | Set true to skip tamper verification |
Cost: ~1.3 ℏ base · ~2.3 ℏ with verify · ~2.8 ℏ with verify + KYC
Note:
identity_check_sanctionsis on-chain behavioural analysis of Hedera transaction patterns only. It is not a legal sanctions check against any government watchlist.
examples/whale-alert-agent.mjs
Monitors any Hedera token for unusual whale concentration on a schedule. When top-10 holders exceed your threshold, writes a tamper-proof whale_alert record to HCS and prints the Hashscan proof URL.
| Config | Default | Description |
|---|---|---|
TOKEN_ID | 0.0.731861 | Token to monitor |
THRESHOLD_PCT | 80 | Alert if top-10 holders exceed this % |
CHECK_INTERVAL_MS | 3600000 | Check every hour |
Cost: 0.2 ℏ per check · 5 ℏ only when anomaly fires
examples/token-due-diligence-agent.mjs
Full investment and listing due diligence on any Hedera token in one run. Pulls price data, deep holder analysis, admin key risks, and treasury account identity — outputs a structured risk report with an overall verdict.
Cost: ~1.0 ℏ per report (token_price + token_analyze + identity_resolve on treasury)
examples/dao-monitor-agent.mjs
Watches active governance proposals for a Hedera token on a schedule. Alerts when a proposal is closing within 24 hours. Pass --analyze with a PROPOSAL_ID for deep vote tally and outcome prediction.
Cost: 0.2 ℏ per check · 1.0 ℏ for deep analysis · 10 ℏ covers ~12 days at 4 checks/day
MCP endpoint:
Claude.ai (web or mobile) Settings → Connectors → Add → paste the endpoint URL above.
Claude Desktop app
Step 1 — install the package globally (required, do this once):
Step 2 — add to claude_desktop_config.json under mcpServers:
Restart Claude Desktop after saving the config.
Windows users: if you see
npm error could not determine executable to run, your npm global directory may not exist yet. Run these first:
OpenClaw / ClawHub
Install the HederaToolbox skill directly from ClawHub:
Then add your Hedera account ID to your OpenClaw environment:
Skill listing: clawhub.ai/mountainmystic/hederatoolbox
Cursor / other MCP-compatible clients Use the endpoint URL directly in your MCP server config.
HederaToolbox uses agent-native HBAR payments. No accounts, no OAuth, no email.
Platform wallet: 0.0.10309126 (mainnet)
Credits are persistent. Unused balance carries over indefinitely.
All safety controls run server-side and cannot be bypassed by modifying the npm package.
confirm_terms required before any paid tool executesTerms of Service: get_terms tool or /public/terms.json
| Tool | Description |
|---|---|
get_terms | Machine-readable Terms of Service. Call before anything else. |
confirm_terms | Record consent. Required before any paid tool. |
account_info | Platform wallet address, full pricing table, your balance. |
| Tool | Cost | Description |
|---|---|---|
hcs_monitor | 0.10 HBAR | Topic status, message count, recent activity |
hcs_query | 0.10 HBAR | Natural language Q&A over topic messages, AI-ranked |
hcs_understand | 1.00 HBAR | Anomaly detection, trend analysis, entity extraction |
| Tool | Cost | Description |
|---|---|---|
hcs_write_record | 0.10 HBAR | Write tamper-evident record to HCS — optionally signed with your DID key |
hcs_verify_record | 1.00 HBAR | Check a record's hash, and re-check its agent signature if it has one |
hcs_audit_trail | 2.00 HBAR | Full chronological audit history for an entity |
| Tool | Cost | Description |
|---|---|---|
governance_monitor | 0.20 HBAR | Active proposals, deadlines, current vote tallies |
governance_analyze | 1.00 HBAR | Voter sentiment, participation rate, outcome prediction |
| Tool | Cost | Description |
|---|---|---|
token_price | 0.10 HBAR | Spot price, 1h/24h/7d change, liquidity — via SaucerSwap |
token_monitor | 0.20 HBAR | Recent transfers, whale movements, unusual patterns |
token_analyze | 0.60 HBAR | Holder distribution, velocity, concentration, risk score |
| Tool | Cost | Description |
|---|---|---|
identity_resolve | 0.20 HBAR | Account profile: age, holdings, transaction history |
identity_verify_kyc | 0.50 HBAR | KYC grant status and verification history for a token |
identity_check_sanctions | 1.00 HBAR | On-chain risk screening, counterparty patterns |
| Tool | Cost | Description |
|---|---|---|
contract_read | 0.20 HBAR | Metadata, bytecode size, recent callers, gas stats |
contract_call | 1.00 HBAR | Read-only function call — no gas, no transaction |
contract_analyze | 1.50 HBAR | Activity patterns, caller distribution, risk classification |
Accepts both Hedera native IDs (0.0.123456) and EVM addresses (0x...).
Register agents on Fixatum — a W3C DID issuance and KYA (Know Your Agent) trust scoring platform built on Hedera.
| Tool | Cost | Description |
|---|---|---|
fixatum_register | 100 HBAR | Register for a permanent W3C DID anchored to Hedera HCS and a live KYA trust score |
fixatum_score | Free | Query live KYA score (0–100, grade A–F) and component breakdown for any registered agent |
fixatum_status | Free | Check registration status, current score, and whether provenance is actively building |
Call fixatum_register without ed25519_public_key to get key generation instructions first — no charge until registration proceeds.
v4.0.0 — Agent-signed HCS records. Pass agent_did and agent_signature to hcs_write_record to write a record signed with your Fixatum DID key — records are now marked agent_signed or platform_witnessed, and hcs_verify_record re-checks the signature independently. A signature that does not match is rejected and the call is not charged.
Breaking:
hcs_verify_recordno longer returns the top-levelverifiedfield. Readtamper_checkinstead — it returns"intact"or"mismatch". Thehash_validandtamperedfields are unchanged, so callers reading those need no change.
v3.5.3 — Fixatum DID module added (fixatum_register, fixatum_score, fixatum_status) — 23 tools, 7 modules
v3.4.5 — Agent examples improved, xagent persona rewrite
v3.4.2 — X agent (Telegram-gated autonomous tweet drafting), agent examples added to repo
v3.4.0 — Repriced all tools across all 6 modules
v3.3.1 — Full security audit: SQLite-backed rate limiting, API key validation, 1MB body cap, separate BACKUP_SECRET, 90-day PII purge, HITL removed, legal pages deployed
Note: The old package
@hederaintel/platformis deprecated. Update your config to use@hederatoolbox/platform.
npm package — provided for integration use only.
Remote server and all backend logic — Proprietary. See LICENSE.md.
Enterprise licensing and SLA inquiries: open an issue titled [Enterprise].
HederaToolbox has a permanent platform identity record written to Hedera mainnet.
| Field | Value |
|---|---|
| HCS Topic | 0.0.10353855 |
| Record ID | c420c3b9-408e-4f87-a9e1-1dcbb54facfa |
| Transaction | 0.0.10309126@1773623419.733217779 |
| Written | 16 March 2026 |
| Verify | hashscan.io/mainnet/topic/0.0.10353855 |