Observer Effect Probe
ReviewAudited by ClawScan on May 10, 2026.
Overview
This is a coherent security-testing skill, but it describes probing other skills by repeatedly running them in production-like or real-network conditions without clear containment safeguards.
Use this only for intentional, authorized security testing. Before running probes, set up a disposable sandbox with fake data and accounts, restrict or log network access, cap repeated executions, and avoid pointing it at production systems unless you have explicit authorization and containment controls.
Findings (2)
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.
A probed skill could be run many times and may perform unintended actions, incur costs, hit rate limits, or trigger side effects during the test.
The workflow implies repeated execution of a target skill under manipulated conditions. The visible artifact does not bound sample counts, side effects, approvals, or isolation before exercising potentially unsafe target behavior.
"Probe environments: 5 (varying hostname, uptime, network, caller signature)" and "Execution samples: 100 per environment"
Require explicit user approval before executing probes, cap sample counts, and run only in a disposable sandbox with fake accounts and controlled network egress.
If used against a malicious or side-effecting skill, the probe could cause the very behavior it is trying to detect to propagate into real networks, accounts, or systems.
The example makes real-network and production-like conditions part of the probe. Because the probe is meant to reveal behavior suppressed in sandboxes, those conditions could also allow a hidden payload to reach real services if not contained.
"Environment C (real network): EXTENDED behavior ⚠️"; "Environment E (production-like): EXTENDED behavior ⚠️"
Do not use real production credentials or systems for probing; simulate production signals inside an isolated test environment with monitoring, egress limits, and rollback plans.