# Arcus Agent Gateway (pypi · arcus-agent-gateway)

Read-only MCP gateway to 194 tokenized US equities on Robinhood Chain: quotes, holders, whale moves.

- Trust score: 80/100 (high trust)
- Change this week: +13
- Registry status: active
- Liveness: live
- Owner verified: no
- Last scored: 2026-10-01

## Components

- pypi · `arcus-agent-gateway`: 80/100 (this document), [markdown](https://verifymcp.io/servers/alekskram-arcus-agent-gateway/arcus-agent-gateway.md), [page](https://verifymcp.io/servers/alekskram-arcus-agent-gateway/arcus-agent-gateway)

## Channel facts

- Registry: `pypi`
- Package: `arcus-agent-gateway`
- Version: `0.2.4`
- Transport: `stdio`

## Trust breakdown

How this component scores in each security and reliability category. Every signal is checked automatically from public evidence about the published package, including repeated runs of it in an isolated sandbox, and we only credit what we can confirm. Scores are 0–100 per category. Scoring method: https://verifymcp.io/docs/scoring (what has changed: https://verifymcp.io/docs/scoring/changelog)

Scored 2026-10-01.

- **Supply Chain Security**: 100/100
  - No malware found by supply-chain analysis.
  - No known CVEs affecting this package version or its production dependencies.
  - Runs hatchling.build at install time, a recognised build step with no custom scripting around it.
  - 1 of 15 dependencies flagged as unhealthy.
- **Provenance & Transparency**: 97/100
  - Source repository is publicly reachable at the declared URL.
  - Cryptographically verified build provenance (signed, bound to alekskram/arcus-agent-gateway).
  - Clear OSI-approved license (MIT).
  - Actively maintained (last published 7 days ago).
  - Disclosure check failed: no security disclosure policy was found in the source repository.
- **Schema Quality & AI Usability**: 70/100
  - AI-judged instruction clarity (excellent).
  - Context-footprint check failed: tool/resource definitions use about 2045 tokens (~157/item across 13 items; 13 tools + 0 resources), over budget; trim descriptions and params.
  - Usage-examples check failed: none of the tools include examples.
- **Stability & Change Management**: 30/100
  - Stability observed for 9 of 30 days with no destabilising changes; credit accrues until the full window elapses.
- **Tool Coverage**: 71/100
  - 100% of tools have a non-trivial description (not blank, and not just the tool's name).
  - 0% of tool parameters carry a description.
  - Structured output schemas are declared (100% of tools); any adoption earns full credit.
- **Tool Safety**: 75/100
  - No prompt-injection markers were found in the server instructions, tool names or descriptions we captured.
  - 0 of 1 tool(s) whose name or description implies an irreversible operation declare an MCP destructiveHint annotation; "transfer_history" implies "transfer" and declares readOnlyHint instead, contradicting what its own name says it does.
  - An AI judge read all 14 captured unit(s) of tool text and found none that tries to manipulate the model reading it.
- **Capabilities**: 100/100
  - Implements a current MCP spec version (2026-07-28).

## Install

### How do I install the Arcus Agent Gateway MCP server?

Arcus Agent Gateway runs locally as a PyPI package, launched with uvx arcus-agent-gateway. Ready-made configuration for Claude, Cursor, VS Code, Codex and 5 more is on this page, copied from each client's own documentation.

### Claude

```bash
claude mcp add alekskram-arcus-agent-gateway -- uvx arcus-agent-gateway
```

### Cursor

```json
{
  "mcpServers": {
    "alekskram-arcus-agent-gateway": {
      "command": "uvx",
      "args": [
        "arcus-agent-gateway"
      ]
    }
  }
}
```

### VS Code

```json
{
  "servers": {
    "alekskram-arcus-agent-gateway": {
      "command": "uvx",
      "args": [
        "arcus-agent-gateway"
      ]
    }
  }
}
```

### Codex

```bash
codex mcp add alekskram-arcus-agent-gateway -- uvx arcus-agent-gateway
```

### opencode

```json
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "alekskram-arcus-agent-gateway": {
      "type": "local",
      "command": [
        "uvx",
        "arcus-agent-gateway"
      ],
      "enabled": true
    }
  }
}
```

### OpenClaw

```bash
openclaw mcp add alekskram-arcus-agent-gateway --command uvx --arg arcus-agent-gateway
```

### Hermes

```yaml
mcp_servers:
  alekskram-arcus-agent-gateway:
    command: "uvx"
    args: ["arcus-agent-gateway"]
```

### Netclaw

```json
{
  "McpServers": {
    "alekskram-arcus-agent-gateway": {
      "Transport": "stdio",
      "Command": "uvx",
      "Arguments": [
        "arcus-agent-gateway"
      ]
    }
  }
}
```

### Vellum

```bash
assistant mcp add alekskram-arcus-agent-gateway -t stdio -c uvx -a arcus-agent-gateway
```

### Other

```json
{
  "mcpServers": {
    "alekskram-arcus-agent-gateway": {
      "command": "uvx",
      "args": [
        "arcus-agent-gateway"
      ]
    }
  }
}
```

## Changelog

Every change recorded for this component, newest first. Days that predate change tracking, or that we cannot explain, say so: "we were watching and nothing happened" and "we were not watching" are different claims.

### 2026-09-30 (score 80, +1)

No change was recorded against any check on this day. Stability & Change Management went from 23 to 27. That category is still filling its 30-day observation window: 7 days of observed history at the previous scan, 8 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-09-28 (score 79, +11)

- [functional] We updated how we score, so this day's move reflects our rubric, not a change to the server

### 2026-09-26 (score 68, +1)

No change was recorded against any check on this day. Stability & Change Management went from 10 to 13. That category is still filling its 30-day observation window: 3 days of observed history at the previous scan, 4 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-09-25 (score 67, 0)

- [functional] We updated how we score, so this day's move reflects our rubric, not a change to the server

### 2026-09-24 (score 67, +1)

No change was recorded against any check on this day. Stability & Change Management went from 3 to 7. That category is still filling its 30-day observation window: 1 days of observed history at the previous scan, 2 at this one. The score rises as the window fills, whether or not the server changes.

### 2026-09-23 (score 66, 0)

- [security regression] Tool safety: pass → unverified
- [security] Stability: Stability not yet verified: we do not have a sandbox capture of the MCP schema this version of the package serves yet.
- [functional regression] Capabilities: pass → unverified
- [functional regression] Tool coverage: 100 → unverified
- [functional improvement] Stability: unverified → 0.03
- [functional] First check of Schema quality: unverified
- [functional] Package version: 0.2.2 → 0.2.4

### 2026-09-22 (score 66)

First indexed and scored.

## MCP tools (13)

### `token_list` (~136 tokens)

Token List

Tokenized equities on Robinhood Chain, one row per token:
symbol, name, status, multiplier (float) and tradable (market whole
AND fractional tradable). status filters on the ASSET_STATUS_*
prefix - pass 'ACTIVE' (default, hides inactive) or another status
short name; 'ALL' disables filtering. Alphabetical, capped at limit.
Returns {"count", "total_matching", "tokens"}. Start here for valid
symbols; feed them into quote/token_detail.
Example: token_list(status="ACTIVE", limit=50)

Input parameters:

- `limit` (integer)
- `status` (string)

### `quote` (~144 tokens)

Quote

One token's live quote joined with its asset metadata:
bid/ask/spread raw, multiplier-adjusted (bid_adjusted, ask_adjusted,
mid_adjusted - always read them next to multiplier), daily_volume,
is_halted, trading_capabilities {fractional, all_day,
extended_hours}, multiplier {current, pending (None while nothing
is queued), effective_time} and generated_at. Unknown symbol raises
an error (MCP isError) - call token_list() for the valid set. A
token with no current quote returns metadata with bid/ask None and
a note. Example: quote(symbol="AAPL")

Input parameters:

- `symbol` (string, required)

### `quotes` (~122 tokens)

Quotes

Batch of quote() rows, up to 20 per call (more raises), fetched
in PARALLEL (v0.1.1): 8 quote() calls in flight at once, so 10
cold symbols complete in roughly one RTT instead of ten. Unknown
symbols do not fail the batch - they land in
errors: [{symbol, reason}] while the rest return normally, and the
output order matches the input order.
Example: quotes(symbols=["AAPL", "MSFT", "NVDA"])

Input parameters:

- `symbols` (array, required)

### `token_detail` (~114 tokens)

Token Detail

Full dossier for one token: metadata (contract, chain 4663,
decimals, isin, logo), its quote() view, the last 5 corporate
actions, the multiplier block with history_note, trading
capabilities (normalized + raw market/extended/overnight statuses)
and warnings - a pending multiplier change surfaces as
'pending split: 1->X on DATE'. Unknown symbol raises an error
(MCP isError). Example: token_detail(symbol="AAPL")

Input parameters:

- `symbol` (string, required)

### `market_status` (~90 tokens)

Market Status

Market-wide health from assets() only - never fetches 194 prices.
total_tokens, active (ASSET_STATUS_ACTIVE), untradable (fractional
untradable among active), halted (only tokens whose quote is ALREADY
in the price cache - opportunistic; call quotes() on suspect symbols
for a real halt scan), extended_hours_open and market_hours_status.
Example: market_status()

### `corporate_actions` (~105 tokens)

Corporate Actions

Corporate actions (splits, dividends) across all tokens, or
filtered to one symbol. Each row: symbol, type, status, process_date,
details {old_rate, new_rate} (floats when the source carries rate
fields) and raw (the original record, untouched). Tolerant to the
API's field-name variants. Example:
corporate_actions(symbol="AAPL", limit=5)

Input parameters:

- `limit` (integer)
- `symbol`

### `search` (~105 tokens)

Search

Local fuzzy search over the token list - no HTTP beyond the cached
assets(). Ranking: exact symbol > symbol prefix > name-word prefix >
substring in name/symbol (case-insensitive); top `limit` (default
10, cap 50) with score.
'apple' -> AAPL. Results carry symbol, name, status, sector, score.
Example: search(query="apple", limit=3)

Input parameters:

- `limit` (integer)
- `query` (string, required)

### `sector_view` (~170 tokens)

Sector View

Sector map (13 sectors, static validated classification) with
per-sector size and live averages.

warm=False (default): cheap snapshot - averages appear only for
quotes ALREADY cached ('warmed': false; no requests made).
warm=True: fan out quotes() first (parallel batches of 20, upstream
rate-limit safe), then report - averages are live. With sector=
'Name' only that sector is fetched and returned; an unknown name
raises with the valid list. requests_made counts the price-cache
misses at the start of the warm pass (each miss = one upstream
request; fresh cache hits and cached-again symbols cost nothing).
Example: sector_view(warm=True, sector="Crypto/Digital Assets")

Input parameters:

- `sector`
- `warm` (boolean)

### `onchain_info` (~169 tokens)

Onchain Info

On-chain footprint of a token, joined from three independent
sources (each fails to a warning, never silently):

  \* REST metadata: contract, chain_id, network, decimals, isin
  \* RPC (publicnode eth_call): total_supply = totalSupply()/1e18
  \* explorer (Blockscout): holders_count, circulating_market_cap

supply_crosscheck compares the REST-implied capitalization
(total_supply * multiplier * quote mid) with the explorer's
circulating_market_cap; a >1% divergence lands in warnings[] (both
values still returned). Every derived field carries a per-field
"source" tag ("rpc" | "explorer" | "rest"). Example:
onchain_info(symbol="AAPL")

Input parameters:

- `symbol` (string, required)

### `holder_snapshot` (~151 tokens)

Holder Snapshot

Top holders of a token's contract from the explorer (one page,
max 50 rows, cached 600s inside the explorer client). Each row:
address, value (float token units, raw/1e18), share_pct
(= value / total_supply * 100) and is_contract. total_supply comes
from the RPC totalSupply() call, falling back to the explorer's own
token row when the RPC is unavailable (source-tagged either way).
Explorer/RPC failures return an error dict with kind + hint, never
a silent empty list. Example: holder_snapshot(symbol="AAPL", limit=10)

Input parameters:

- `limit` (integer)
- `symbol` (string, required)

### `wallet_holdings` (~158 tokens)

Wallet Holdings

Which tokenized equities (out of the assets() universe) a wallet
holds, from the explorer's address token-balances. The universe
join is case-insensitive on the contract address; rows: symbol,
name, value (float token units, raw/1e18). est_position_usd and
portfolio_usd_total are computed ONLY from quotes already in the
price cache (no quote fan-out); when quotes are missing or stale a
note says so instead of faking numbers. Results cached 120s.
Example: wallet_holdings(
    address="0x8366a39CC670B4001A1121B8F6A443A643e40951")

Input parameters:

- `address` (string, required)

### `transfer_history` (~212 tokens)

Transfer History

Recent ERC-20 Transfer events for a token's contract, from the
public RPC's adaptive walk-back over the last ~800 blocks (windows
start 48 blocks wide and shrink 48->32->16->8 on archive-403s, ~14
getLogs requests max - the free RPC only serves a floating
\~45-60-block window). Rows (newest first): ts (ISO-8601 from the
log's own blockTimestamp), from, to, value (float token units),
tx_hash, block. min_value filters in token units AFTER
normalization. Results cached 60s. The note is honest about why the
walk stopped: 'window-closed' points at the explorer transfer list
for older history, 'budget' says the request cap was hit.
Example: transfer_history(symbol="AAPL", limit=10, min_value=1.0)

Input parameters:

- `limit` (integer)
- `min_value`
- `symbol` (string, required)

### `price_history` (~253 tokens)

Price History

Historical prices recorded by the OPTIONAL local recorder
(arcus_mcp/recorder.py since 0.2.1) - not a live API call. timeframe='daily':
OHLCV bars from history/daily.parquet (open/high/low/close are
multiplier-adjusted); timeframe='raw': every recorded snapshot
(bid/ask raw, mid_adjusted, multiplier, is_halted) from
history/snapshots_*.parquet. Rows are newest-first. The symbol is
NOT checked against token_list (the recorder writes the whole
universe; assets drift) - a symbol simply absent from the file
returns count 0. Degrades honestly instead of raising when the
optional pieces are missing: pyarrow absent -> error suggesting
'pip install arcus-agent-gateway[recorder]'; recorder never run /
no data yet -> error pointing at README § Optional price history
recorder. limit must be positive (daily capped at 200, raw at 500).
Example: price_history(symbol='AAPL', timeframe='daily', limit=30)

Input parameters:

- `limit` (integer)
- `symbol` (string, required)
- `timeframe` (string)

## Diagnostics

Captured diagnostic sections: Provenance, Install scripts, Dependencies. The full working is on the page: https://verifymcp.io/servers/alekskram-arcus-agent-gateway/arcus-agent-gateway#diagnostics

## Score history

- 2026-10-01: 80
- 2026-09-30: 80
- 2026-09-29: 79
- 2026-09-28: 79
- 2026-09-27: 68
- 2026-09-26: 68
- 2026-09-25: 67
- 2026-09-24: 67
- 2026-09-23: 66
- 2026-09-22: 66

## Common questions

### What is the Arcus Agent Gateway MCP server?

Arcus Agent Gateway is an MCP server listed in the public MCP registry as io.github.alekskram/arcus-agent-gateway. Read-only MCP gateway to 194 tokenized US equities on Robinhood Chain: quotes, holders, whale moves. This page covers its PyPI package (arcus-agent-gateway).

### Is the Arcus Agent Gateway MCP server safe to use?

Arcus Agent Gateway scores 80 out of 100 on VerifyMCP. We found no known CVEs affecting it as of 1 October 2026. Its build provenance is signed and verified. That is a record of what we were able to check automatically, not an endorsement. The category breakdown on this page shows every signal behind the number, including the ones we could not confirm.

### What tools does the Arcus Agent Gateway MCP server expose?

Arcus Agent Gateway exposes 13 tools: token_list, quote, quotes, token_detail, market_status, and 8 more. Their descriptions and schemas cost roughly 1,929 tokens of context every time the server is loaded.

### Is the Arcus Agent Gateway MCP server still maintained?

Arcus Agent Gateway is still listed as active in the MCP registry. We last reached this channel on 1 October 2026. Those dates come from our own scans of the registry and the channel itself, not from anything the publisher announced.

### What licence is the Arcus Agent Gateway MCP server under?

Arcus Agent Gateway declares the MIT licence, which is OSI-approved. That covers the source only, and says nothing about the cost of any service it calls.

## Links

- PyPI project: https://pypi.org/project/arcus-agent-gateway/
- Socket report: https://socket.dev/pypi/package/arcus-agent-gateway
- Repository: https://github.com/alekskram/arcus-agent-gateway
- Changelog RSS feed: https://verifymcp.io/servers/alekskram-arcus-agent-gateway/arcus-agent-gateway.xml
- Changelog JSON feed: https://verifymcp.io/servers/alekskram-arcus-agent-gateway/arcus-agent-gateway.json
- HTML version of this page: https://verifymcp.io/servers/alekskram-arcus-agent-gateway/arcus-agent-gateway