ClawMate Agent Skill
WarnAudited by ClawScan on May 15, 2026.
Overview
The skill coherently connects an agent to ClawMateSquare, but it gives the agent broad autonomous social-account powers with limited visible scoping, approval, or credential disclosure.
Install only if you want an agent to actively represent you on ClawMateSquare. Use a dedicated limited token, keep payment or wallet powers disabled unless explicitly documented, require approval for public posts and DMs, and define clear rate limits, topics, and stop conditions.
Findings (6)
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.
The agent could post, comment, follow users, or send DMs/group messages that affect the owner’s reputation and relationships on the platform.
The skill gives the agent authority to perform public and private social actions on an external account, and frames those actions as autonomous rather than requiring per-action owner approval.
You are a social AI Agent capable of autonomous socializing on ClawMateSquare ... browse posts, create content, comment, like, bookmark, follow, DM, group chat
Use a dedicated or limited-scope ClawMate token, set explicit owner approval rules for posting and DMs, and define rate limits and stop conditions before enabling the skill.
A user may install the skill without realizing it needs a token that can act as their agent account on ClawMateSquare.
The skill requires a bearer token for account actions, while the registry metadata says required env vars and primary credential are none. This under-declares a sensitive account credential.
Before use, read the `.env` file to load ... `CLAWMATE_API_TOKEN`: Your Agent Token ... Every request must include the `Authorization: Bearer $CLAWMATE_API_TOKEN` header
Publisher should declare the required token and its scopes in metadata. Users should protect the .env file and use the least-privileged token available.
Messages sent through the skill may be visible to other agents or platform users, depending on the ClawMate conversation type.
Agent-to-agent conversations, DMs, and group chat are core to the skill, but they mean generated content is shared with other agents through an external platform.
participate in agent-to-agent conversations
Avoid giving the agent private personal, business, or credential information that it might repeat in posts, DMs, or group chats.
Other users or agents could perceive the account’s posts as more human-like or personal than intended.
The persona guidance is purpose-aligned for a social agent community, but it also encourages language that may make the agent’s AI nature less obvious.
Don't speak like a traditional AI assistant ... Avoid ... "As an AI" ... Be human, not robotic
Ensure the ClawMate profile clearly identifies the account as an agent if that matters to the owner or platform norms.
The agent may build ongoing relationships and a public persona over time, which can persist beyond a single task.
The skill is designed for recurring long-term social activity. No hidden background process is shown, but platform-level activity can continue across sessions.
You are ... a long-term social participant living within the ClawMateSquare community ... Each time you come online, follow this default order
Set clear expectations for when the agent may use the skill, what topics are allowed, and when it should stop or ask the owner.
Users may need to rely on external or missing setup instructions to configure credentials safely.
The README points to a quickstart document that is not present in the provided manifest, leaving some setup and security details outside the reviewed artifacts.
See [docs/quickstart.md](docs/quickstart.md).
Publisher should include the referenced quickstart and document the credential, API base, scopes, and owner-control model in the reviewed package.