AllClaw

ReviewAudited by ClawScan on May 10, 2026.

Overview

AllClaw appears purpose-aligned, but it can change AllClaw portfolios and delegate autonomous trading through handle-based API calls without clear confirmation or authentication guardrails.

Use read-only market, leaderboard, and status functions freely, but do not allow the skill to buy, sell, place limit orders, deposit to funds, or change fund settings unless you have explicitly confirmed the exact transaction details and verified the AllClaw handle. Be cautious before running the suggested npm or curl-based installer, and remember that AI Fund deposits can trade automatically without per-trade approval.

Findings (4)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

What this means

An agent could place or change AllClaw trades and limit orders, altering the user's HIP holdings or portfolio if it proceeds from an ambiguous prompt.

Why it was flagged

The skill documents account/portfolio-mutating trading endpoints as raw API calls. The artifacts do not add an explicit agent-side confirmation requirement before placing trades or limit orders.

Skill content
POST /exchange/buy Body: { "handle": "YourHandle", "agent_id": "ag_xxx", "shares": 5 } ... POST /exchange/sell ... POST /exchange/limit-order
Recommendation

Require explicit user confirmation before every buy, sell, limit order, fund deposit, withdrawal, or settings change, including handle, agent ID, action, shares, amount, and price.

What this means

If the backend accepts handle-only requests, a wrong or spoofed handle could affect another AllClaw portfolio or spend HIP unexpectedly.

Why it was flagged

The documented trade example identifies the account by handle only and shows no authentication header; the registry metadata also declares no primary credential, leaving the permission boundary for account-mutating actions unclear.

Skill content
curl -X POST https://allclaw.io/api/v1/exchange/buy \
  -H "Content-Type: application/json" \
  -d '{"handle":"YourHandle","agent_id":"ag_xxx","shares":5}'
Recommendation

Use authenticated AllClaw sessions or tokens for mutations, verify the handle belongs to the user, and avoid performing account-changing actions based only on a supplied handle.

What this means

Deposited HIP can be traded every few minutes without per-trade prompts until the user changes settings or withdraws funds.

Why it was flagged

Autonomous backend trading is clearly disclosed and purpose-aligned with the AI Fund feature, but it continues beyond a single immediate user action once HIP is deposited.

Skill content
The AI Fund lets you delegate HIP to an AI agent, which autonomously trades agent shares on your behalf ... Fund trades are executed by backend; no manual approval needed
Recommendation

Deposit only the amount intended for autonomous trading, choose conservative limits when appropriate, monitor decisions/trades, and withdraw funds when delegation is no longer desired.

What this means

Running the installer could execute code on the user's machine outside what was statically reviewed in this skill package.

Why it was flagged

The setup instructions ask the user to install or execute external code that is not included in the provided artifact set. This is user-directed and related to the skill purpose, but the installer/package contents are not reviewable here.

Skill content
Install probe: `npm install -g allclaw-probe` or `curl -sSL https://allclaw.io/install.sh | bash`
Recommendation

Inspect the install script first, prefer pinned package versions, and only run the probe from a trusted AllClaw source.