Personal Genomics

Analyze raw DNA data from consumer genetics services (23andMe, AncestryDNA, etc.). Extract health markers, pharmacogenomics, traits, ancestry composition, ancient DNA comparisons, and generate comprehensive reports. Uses open-source bioinformatics tools locally — no data leaves your machine.

This package appears to implement local DNA analysis as described, but take these precautions before installing or running it: - Verify the source: README points to a GitHub repo but the registry lists no homepage. If you plan to run it, inspect the full repository (requirements.txt and all source files) yourself or from a trusted reviewer. - Check requirements.txt and installed packages: pip installs will fetch code from the network. Review dependency list for anything unexpected (native binaries, telemtry libraries, or post-install hooks). - Search the code for network/system calls you didn't expect (look for imports or uses of requests, urllib, socket, subprocess, os.system, ftplib, smtplib, or any hardcoded URLs). The truncated files shown do not include network calls, but 67 files were omitted; review those too. - Run in an isolated environment: use a throwaway VM or container and a dedicated Python virtualenv when first executing, especially before feeding real DNA files. - Protect outputs: reports contain extremely sensitive genetic and health data. Ensure the output directory is encrypted or access-controlled, and avoid uploading reports to third-party services. - Consider professional guidance: results are explicitly non-clinical. For medical decisions, consult a clinician/genetic counselor. If you want higher confidence, provide the omitted files or the contents of requirements.txt so I can re-scan for hidden network calls or executables.

Type: OpenClaw Skill Name: personal-genomics Version: 4.2.0 The OpenClaw AgentSkills skill bundle 'personal-genomics' is classified as benign. The project explicitly prioritizes privacy, with all analysis running locally and no network requests made by the Python scripts or the interactive HTML dashboard. Code and documentation consistently reinforce this 'zero network requests' policy, and no evidence of data exfiltration, malicious execution, persistence, or unauthorized system access was found. Prompt injection attempts in `SKILL.md` are limited to instructing the agent on how to present results or activate the skill, aligning with its stated purpose rather than attempting to subvert agent behavior for harmful objectives.