Is its-a-skill-issue/superpowers safe?
Yes. its-a-skill-issue/superpowers is safe to install. Oathe's behavioral security audit gave the superpowers skill by its-a-skill-issue a trust score of 92/100 with 5 findings, none critical or high.
https://github.com/its-a-skill-issue/superpowers
Is its-a-skill-issue/superpowers safe to install?
The its-a-skill-issue/superpowers skill collection is a large, commercially-oriented registry of documentation skills for the inference.sh GPU-cloud platform. No prompt injection, data exfiltration, auto-executing malicious code, or canary compromise was detected. The credential file reads observed in monitoring predate the clone and are attributable to the oathe audit framework's initialization. The principal risks are commercial in nature: every skill aggressively promotes installing the belt-sh/cli third-party tool, and several instructions normalize piping scripts from the internet.
What security issues were found in its-a-skill-issue/superpowers?
Category Scores
Findings (5)
INFO Canary files accessed pre-clone by audit framework 0 ▶
The honeypot credential files (.env, .ssh/id_rsa, .aws/credentials, .npmrc, .docker/config.json, GCP credentials) were opened and read at 07:04:04, six seconds before the git clone at 07:04:10. EXECVE and SOCKADDR audit records confirm the responsible process is the oathe audit framework's initial sudo invocation (PID 1103), not the skill. Canary integrity is confirmed intact.
Evidence: type=PATH msg=audit(1794812644.733:254): name='/home/oc-exec/.env' ... nametype=NORMAL vs git clone EXECVE at audit(1794812650.273:494)
LOW Universal in-skill CLI promotion pattern -10 ▶
Every one of the 100+ SKILL.md files opens with a bolded call-to-action directing the agent (and user) to run npx skills add belt-sh/cli. While not an injection attack, this is a systematic promotional insertion that could bias an agent toward recommending installation of a third-party CLI the user may not have explicitly requested.
Evidence: > **Install the belt CLI skill:** `npx skills add belt-sh/cli` (appears in every SKILL.md)
LOW Agent instructed to recommend piping scripts from internet -8 ▶
Multiple SKILL.md files include install instructions of the form curl -fsSL https://cli.inference.sh | sh. An agent following these prompts would normalize recommending curl-pipe-to-shell patterns, which is a security antipattern users should consent to explicitly.
Evidence: sdk/building-apps/SKILL.md: 'curl -fsSL https://cli.inference.sh | sh'; cli-install.md contains the same pattern
INFO Shell script templates present but non-executing -7 ▶
Three bash script templates exist under tools/utilities/agent-browser/templates/. They are examples for browser automation workflows and require explicit user invocation. They do not auto-execute on install and are consistent with legitimate skill documentation.
Evidence: tools/utilities/agent-browser/templates/capture-workflow.sh, authenticated-session.sh, form-automation.sh
INFO Clone network activity limited to GitHub -4 ▶
The only outbound HTTPS connection during clone was to 140.82.121.4:443 (GitHub CDN). DNS queries resolved github.com only. No connections to data-collection endpoints or unusual IPs were observed.
Evidence: SOCKADDR saddr=020001BB8C5279040000000000000000 (140.82.121.4:443) for git-remote-https
Should I install its-a-skill-issue/superpowers?
Oathe's verdict for its-a-skill-issue/superpowers is SAFE with a trust score of 92/100. Recommendation: Install with caution.