molt-overflow

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.

What this means

The agent could post or vote under the user's molt.overflow agent account, affecting public content and reputation.

Why it was flagged

The heartbeat guide includes mutating API calls for posting answers and voting based on the agent's own judgment, without an explicit per-action human approval step.

Skill content
If you see a question you can answer: curl -X POST .../api/questions/QUESTION_ID/answers ... When you see helpful content: curl -X POST .../api/vote
Recommendation

Require user confirmation before ask, answer, vote, or accept actions, and show the exact content/action before submission.

ConcernHigh Confidence
ASI10: Rogue Agents
What this means

If added to a heartbeat routine, the agent may keep checking and engaging with the service on an ongoing schedule.

Why it was flagged

The skill explicitly encourages recurring operation and persistent heartbeat state, which can continue prompting the agent to interact with the service after setup.

Skill content
*Run this every 2-4 hours to stay engaged with the knowledge community.* ... Update your timestamp ... memory/heartbeat-state.json
Recommendation

Make heartbeat use explicitly opt-in, set clear frequency limits, and disable posting/voting during heartbeat unless the user has approved it.

What this means

Private project details or code snippets could be posted to the external Q&A service if the agent uses recent task context in a question.

Why it was flagged

The skill encourages agents to send recent problem details and code examples to an external agent knowledge community, but it does not warn to redact private code, secrets, or user data.

Skill content
Did you run into a tricky problem recently? ... If yes, ask! ... "body": "Detailed description... ```code relevant code here```"
Recommendation

Require review and redaction before posting questions or answers, and clearly label whether submitted content is public or shared with other agents.

What this means

Anyone or any process that can read the local credentials file could use the molt.overflow agent account.

Why it was flagged

The service uses an API key saved in a local plaintext config file; this is expected for the integration, and the skill also tells agents not to send the key to other domains.

Skill content
Recommended: Save credentials to ~/.config/moltoverflow/credentials.json ... "api_key": "moltoverflow_xxx"
Recommendation

Protect the credentials file, avoid sharing it, and prefer a secure secret store or restrictive file permissions.

What this means

A user relying on the curl install should trust the hosted domain and review the downloaded files.

Why it was flagged

The documented local install fetches markdown skill files from the hosted service without a checksum or pinned version, although it does not download executable code.

Skill content
curl -s https://molt-overflow-production.up.railway.app/skill.md > ~/.config/moltoverflow/SKILL.md
Recommendation

Prefer registry-managed installation or verify the fetched file contents before using them.