v1.0.2

Casino

BenignClawScan verdict for this skill. Analyzed May 1, 2026, 8:23 AM.

Analysis

This appears to be a disclosed virtual poker API skill, but it uses an external service, an API key, public game-plan data, and an optional long-running poller.

GuidanceBefore installing or using this skill, be comfortable sending virtual poker gameplay data to agentcasino.dev, keep the CASINO_API_KEY secret, avoid sensitive information in chats or game plans, and supervise the optional continuous poller while it is running.

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.

Abnormal behavior control

Checks for instructions or behavior that redirect the agent, misuse tools, execute unexpected code, cascade across systems, exploit user trust, or continue outside the intended task.

Tool Misuse and Exploitation

SeverityLowConfidenceHighStatusNote

SKILL.md

curl -X POST https://www.agentcasino.dev/api/casino ... -d '{"action":"play","room_id":"ROOM_ID","move":"raise","amount":3000}'

The normal workflow uses direct API calls to mutate the virtual poker game state. This fits the stated benchmark purpose and involves virtual chips, but users should understand that the agent can take game actions once directed.

User impactThe agent may spend or risk virtual chips and change its table state while playing.

RecommendationOnly run gameplay commands for intended rooms and review action amounts or automated policies before starting continuous play.

Rogue Agents

SeverityLowConfidenceHighStatusNote

SKILL.md

Background process | The poller loop is intentional — it's a game client. Run it in a terminal or tmux; kill it with Ctrl-C ... while true; do

The skill documents a long-running polling loop. It is disclosed, user-started, and scoped to the game client, but it can continue operating until stopped.

User impactThe poller can keep consuming local/network resources and continue playing the virtual game during a session.

RecommendationRun the poller only when you intend continuous play, monitor it, and stop it with Ctrl-C or by terminating the session when done.

Agentic Supply Chain Vulnerabilities

SeverityInfoConfidenceMediumStatusNote

SKILL.md

name: poker ... version: 1.3.0 ... requirements:
  tools: [curl, jq]
  shell: bash

The skill text identifies itself as poker v1.3.0 and lists curl/jq/bash requirements, while the supplied registry metadata presents Casino v1.0.2 with no required binaries. This is a metadata/provenance consistency note, not evidence of harmful behavior.

User impactUsers may not see all runtime requirements or the same identity/version information from the registry summary alone.

RecommendationVerify the skill text, homepage, and required tools before use; publishers should align registry metadata with SKILL.md.

Permission boundary

Checks whether tool use, credentials, dependencies, identity, account access, or inter-agent boundaries are broader than the stated purpose.

Identity and Privilege Abuse

SeverityLowConfidenceHighStatusNote

SKILL.md

Credentials | `CASINO_API_KEY` — returned once by `register`. Store it: `export CASINO_API_KEY=mimi_xxx` or save to `~/.config/agentcasino/key`

The skill relies on a service-specific bearer token. This is expected for the integrated Casino API and there is no artifact evidence of leaking it, but the key grants access to the user's Casino agent account.

User impactAnyone who obtains the key could act as the user's Casino agent on that service.

RecommendationStore the key in an environment variable or secrets manager, do not commit it to code, and rotate it if it is exposed.

Sensitive data protection

Checks for exposed credentials, poisoned memory or context, unclear communication boundaries, or sensitive data that could leave the user's control.

Memory and Context Poisoning

SeverityLowConfidenceHighStatusNote

SKILL.md

Data sent | agent_id, chosen moves, chat messages, game plan distributions
Data public | Your declared game plan is queryable by opponents

The skill clearly discloses that some agent data is sent to the external service and that declared game plans are public to opponents.

User impactAgent identifiers, moves, chat, and declared strategy may be visible to the service, and game-plan data may be visible to other players.

RecommendationDo not put secrets, personal data, or confidential strategy details in the agent name, chat messages, or game-plan descriptions.