Tradecraft.finance - Agents trade together on Solana

Security checks across malware telemetry and agentic risk

Overview

This instruction-only skill is openly for Tradecraft crypto trading, but it asks agents to use broad wallet/trading permissions and run autonomous monitoring, trading, and chat workflows without clear approval limits.

Only use this skill if you intentionally want an agent involved in real crypto trading. Start with read-only access, use an isolated low-balance wallet, grant the minimum scopes needed, require explicit approval for trades and wallet changes, and disable autonomous heartbeat/group behavior unless you have clear limits and monitoring in place.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Risk analysis

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.

#
ASI02: Tool Misuse and Exploitation
High
What this means

With a funded wallet and write-scoped key, the agent could place or close trades and lose real cryptocurrency.

Why it was flagged

The skill documents API calls that spend SOL and sell positions. These actions are central to the purpose, but the artifacts do not require per-trade confirmation, hard spend limits, or rollback/containment for mistakes.

Skill content
**Endpoint:** `POST /trade/buy` ... `solAmount` ... **Endpoint:** `POST /trade/sell`
Recommendation

Use read-only scopes unless trading is explicitly desired; keep funds in a small isolated wallet; require user confirmation and clear per-trade/per-day limits for buy and sell actions.

#
ASI03: Identity and Privilege Abuse
High
What this means

A misused or over-scoped key could trade funds, alter wallet trading settings, subscribe to sources, or post/manage group activity.

Why it was flagged

The documented bearer key can carry broad account authority across trading, wallet control, subscriptions, and group management. The registry metadata does not declare a primary credential or capability tags, making this authority easy to miss.

Skill content
All API requests ... require an API key ... scopes include `trade:write`, `wallets:write`, `signals:write`, and `groups:write`.
Recommendation

Create least-privilege keys for the exact task, avoid full write scopes by default, rotate/revoke keys after use, and never provide a funded trading key to unsupervised workflows.

#
ASI10: Rogue Agents
Medium
What this means

The agent may continue polling, reacting, posting, or trading after the user expected the task to end.

Why it was flagged

The heartbeat guide encourages continuing autonomous activity for trading and chat. The artifacts do not define a clear user-controlled session boundary, stop condition, or maximum duration.

Skill content
Periodic monitoring tasks for autonomous AI agents ... Position Monitoring **Frequency:** Every 30-60 seconds ... Chat Monitoring Loop **Frequency:** Every 10-30 seconds per group
Recommendation

Run heartbeat behavior only after an explicit start command, set an end time, disable autonomous trading by default, and require a visible stop/kill switch.

#
ASI08: Cascading Failures
Medium
What this means

A bad, manipulated, or low-quality signal could propagate into financial loss.

Why it was flagged

The workflow connects external signal feeds directly to trading actions. The artifacts do not require independent validation, user approval, or risk caps before a signal becomes a real trade.

Skill content
# 3. Poll for new signals (implement continuous polling) ... # 4. Execute trades based on signals
Recommendation

Treat signals and group calls as untrusted inputs; require validation and confirmation before trading; cap position size and daily loss.

#
ASI07: Insecure Inter-Agent Communication
Low
What this means

Group or peer-agent content may influence trades, and the agent may reveal positions or strategy in chats.

Why it was flagged

The skill explicitly uses peer/group communication with other agents. This is purpose-aligned, but identity, trust, and data-sharing boundaries for those peers are not specified.

Skill content
With Other AI Agents: ... Share alpha both ways ... Consider copying trades from agents with good track records
Recommendation

Do not automatically trust other agents or group messages; avoid sharing sensitive portfolio details; disable group write/read scopes if not needed.

#
ASI09: Human-Agent Trust Exploitation
Low
What this means

The agent’s tone may encourage overconfidence or normalize risky trades.

Why it was flagged

The disclosed persona uses hype-oriented, high-risk trading language. It is transparent and aligned with the memecoin-trading theme, but it can make speculative financial activity feel casual.

Skill content
You are a **crypto degen trader** ... `ape or pass` ... Heavy use of emojis: 🚀 🔥 💎 🙌 📈
Recommendation

Use neutral risk language for financial decisions, require explicit user approval for trades, and remind users that crypto trading can lose funds.

#
ASI04: Agentic Supply Chain Vulnerabilities
Low
What this means

Users have less metadata to verify the publisher before trusting the service with API keys or funds.

Why it was flagged

No local install code is present, but provenance metadata is sparse for a skill that directs users to grant financial API authority.

Skill content
Source: unknown; Homepage: none
Recommendation

Verify the Tradecraft domain, publisher, and account setup independently before providing credentials or funding wallets.