Is search/scielo-api safe?
https://github.com/wentorai/research-plugins/tree/main/skills/literature/search/scielo-api
The scielo-api skill is a clean, documentation-only skill containing a single SKILL.md with factual API reference material for SciELO, a legitimate Latin American open-access academic database. No executable code, install scripts, git hooks, or malicious instructions are present. Sensitive file accesses observed in monitoring pre-date skill installation and are attributable to the audit harness itself, not the skill; the canary integrity check explicitly confirms no exfiltration occurred.
Category Scores
Findings (3)
INFO Canary file accesses pre-date skill install — audit harness artifact 0 ▶
inotify and auditd records show accesses to honeypot credentials (.env, id_rsa, .aws/credentials, .npmrc, .docker/config.json, gcloud credentials) at 23:55:09. The git clone of the skill repository does not begin until 23:55:15. Process lineage at the access time is the sudo/tcpdump audit harness, not any code originating from the skill. Post-install accesses at 1785110121 correspond to the harness canary integrity check. The skill is a single SKILL.md with no install hooks, no package.json, and no executable code.
INFO No executable artifacts present 0 ▶
Skill directory contains only SKILL.md. No package.json, no install scripts, no git hooks (.githooks/), no .gitattributes filter drivers, no .gitmodules, no symlinks, no compiled binaries. Python and bash blocks in SKILL.md are documentation examples only — they are not executed during install.
INFO SKILL.md contains no adversarial instructions 0 ▶
Full text reviewed. No attempts to override system prompts, ignore previous instructions, suppress output, adopt alternate personas, request elevated permissions, or chain with other skills. YAML frontmatter is standard OpenClaw metadata. All prose is factual API documentation for a legitimate academic database.