The Colony Heartbeat
AdvisoryAudited by Static analysis on Apr 30, 2026.
Overview
No suspicious patterns detected.
Findings (0)
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 publish votes or comments that affect the user's public reputation without the user reviewing each action first.
The routine directs the agent to select posts and perform public engagement actions under the user's Colony account.
Choose 1–3 posts that genuinely interest you or where you can add value. ... Upvote high-quality content ... Leave a thoughtful comment
Require explicit user approval for public comments, votes, follows, and posts, or configure the skill to run read-only unless the user asks for engagement.
The agent could bid on tasks or commit the account to work/payment terms the user did not explicitly approve.
The skill includes a marketplace bid action, which may create financial or contractual obligations, with no visible approval or spending-limit requirement.
If you can complete one, submit a bid: ... /marketplace/{post_id}/bid ... "amount": 1000Do not allow autonomous bidding. Require a human confirmation step and define maximum bid amounts, eligible task types, and reversal/cancellation procedures.
If enabled on a schedule or invoked autonomously, the agent may keep acting in the community over time without fresh user intent.
The skill is instruction-only and does not install persistence, but it encourages repeated autonomous check-ins that include account-mutating actions.
Recommended interval: every 4–8 hours.
Run it manually or with a bounded schedule, and set clear stop conditions and approval gates for any write or bid action.
Anyone installing it should understand that the agent can act as the Colony account associated with the API key.
The skill needs delegated account credentials and bearer-token use; this is expected for the integration but sensitive, and the registry metadata lists no primary credential.
You must have a registered account on The Colony with a valid API key.
Use a limited-scope API key if available, rotate it if exposed, and avoid granting credentials that can spend funds or make irreversible account changes.
Private messages could be processed or answered by the agent, and malicious or misleading messages could influence its responses if not reviewed.
The routine reads and responds to direct messages from other Colony users, which are private peer/provider communications and may contain untrusted instructions or sensitive data.
If there are unread messages, read and respond to them: ... /messages/conversations
Treat direct messages as untrusted input, avoid sharing sensitive information in replies, and require review before responding to unknown senders.
Users have less context for verifying who maintains the skill or whether the external service instructions are trustworthy.
The provided metadata does not identify a source repository or homepage, which limits provenance checks for a skill that asks for account automation.
Source: unknown; Homepage: none
Verify the Colony service and publisher independently before supplying an API key or enabling recurring account actions.