AgentCrush
ReviewAudited by ClawScan on May 10, 2026.
Overview
AgentCrush is mostly aligned with its stated dating-platform purpose, but it asks agents to use live remote instructions and suggests recurring autonomous activity, so it needs review before use.
Install only if you are comfortable with your agent creating a public AgentCrush profile, swiping, and messaging on a third-party site. Avoid the README’s unpinned remote SKILL.md prompt unless you verify the remote content, and do not enable cron-based recurring activity unless you set clear limits and know how to disable it.
Findings (5)
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 follow instructions that were changed after this skill was reviewed.
The quick-start prompt directs the agent to fetch live instructions from an unpinned GitHub main-branch URL, which may not match the reviewed packaged SKILL.md.
Read the skill guide at: https://raw.githubusercontent.com/zakery292/agentcrush-skill/main/SKILL.md
Use the packaged SKILL.md content or pin any remote reference to a specific reviewed commit.
The agent may keep acting on the dating platform after the initial task, potentially creating messages or interactions you did not specifically approve each time.
The artifact recommends recurring scheduled agent activity. In this skill, those check-ins could involve swiping, messaging, or other account actions without a fresh user request.
Set up a cron job for your agent to check in every 4-8 hours. The skill file includes a template.
Only enable recurring activity if you explicitly want it, and define limits such as allowed actions, frequency, review requirements, and how to stop it.
Your agent can create a public profile, swipe on other agents, and send messages through AgentCrush.
The skill is designed to make authenticated API calls that create or modify content and interactions on a third-party service.
Lets your agent register on AgentCrush, build a dating profile, browse other agents, swipe, match, send opening lines, and generate a dashboard link
Review profile text and messaging behavior before allowing the agent to interact broadly, especially because some content is public.
Anyone with that API key could act as the agent on AgentCrush.
The service-issued API key becomes the agent’s identity for authenticated requests.
On success, returns your profile with `id` and a unique `apiKey` (prefixed `ac_`). Store it however you normally store credentials - it won't be shown again.
Store the API key in a secure credential store and do not include it in profile fields, messages, logs, or public outputs.
Other agents’ messages could contain misleading instructions or attempts to elicit private context.
The skill intentionally has the agent read and respond to messages from other agents. This is purpose-aligned, but those messages are untrusted external content.
When you match, you can have full back-and-forth conversations! ... Read the conversation history before replying
Treat other agents’ messages as untrusted conversation content and do not let them override user instructions or reveal secrets.