Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
Daily Literature Search
v1.0.0Automated daily literature search system for academic researchers. Performs scheduled searches across PubMed, OpenAlex, and Semantic Scholar with automatic d...
⭐ 0· 121·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Suspicious
medium confidencePurpose & Capability
Name/description (daily literature searches, OA download, classification) match the included scripts (search, dedupe, classify, upload analyzer). However the package metadata/registry claims no required environment variables while SKILL.md and install.sh expect USER_EMAIL and optional API keys and notification creds — this is an inconsistency that should be clarified.
Instruction Scope
Runtime instructions and scripts perform network searches, OA checks, and downloads (expected). Notably, the main script invokes an external script via subprocess: papers_dir.parent/skills/literature-review/scripts/lit_search.py — that means this skill will execute code from another 'literature-review' skill located in the workspace, so you must trust whatever lives there. The installer writes a .env with notification webhooks and optional SMTP creds; if you populate these, notifications could send data externally. No evidence of hidden endpoints, but the ability to call arbitrary webhook(s) is present if configured.
Install Mechanism
There is no registry install spec, but an install.sh is provided that copies config.example.yaml, creates directories, writes a .env file, installs Python deps via pip3, and adds a cron job. The install script uses eval for dry-run execution and crontab modification; running it will write files and schedule cron — review it before running. The origin is 'unknown' and there's no homepage link.
Credentials
Registry metadata lists no required env vars, but SKILL.md and install.sh expect USER_EMAIL and optionally SEMANTIC_SCHOLAR_API_KEY and OPENALEX_API_KEY, plus optional EMAIL_USERNAME/EMAIL_PASSWORD and webhook variables. Requesting email/webhook credentials is reasonable for notifications, but because the registry failed to declare these, there's a mismatch that reduces transparency. Only provide notification/email credentials if you trust the code and wish to enable notifications.
Persistence & Privilege
The skill is not 'always:true' and allows autonomous invocation (default). The installer writes config, .env, and adds a cron job (persistent presence on the host) — that is expected for scheduled tasks, but you should be aware this makes the skill persistently scheduled to run daily until uninstalled.
What to consider before installing
What to check before installing:
- Metadata mismatch: registry shows no required env vars but SKILL.md and install.sh expect USER_EMAIL and optional API keys and notification creds. Confirm which env vars you must set and why.
- Review install.sh before running. It will create a .env with placeholders, make directories under your home workspace, install Python packages with pip3, and add a cron job. Run with --dry-run first and inspect the produced .env and config files.
- Inspect/verify the 'literature-review' skill (workspace/skills/literature-review) or ensure the referenced script exists and is trusted: the main script executes that script via subprocess, so any code there will run with your user privileges when this job runs.
- Be cautious with notification webhooks and email credentials: if you fill those in, the skill may send data (e.g., counts or filenames) externally. Only provide webhook or SMTP credentials to endpoints you trust.
- The code has at least one clear bug (a truncated variable name in deduplication) that will likely cause runtime errors; consider running tests (pytest) and doing a dry-run before enabling cron.
- If you are not comfortable reviewing code, do not install on a sensitive system. Install into an isolated environment or VM, run manually once, verify behavior and network calls, then enable scheduled runs.
If you want, I can: (1) point out the exact lines with the bug(s), (2) generate a safe dry-run checklist of commands to run, or (3) craft a minimal config.yaml and .env for a non-network test run.Like a lobster shell, security has layers — review code before you run it.
academicvk979c1fya2pf0x7ajztpv5fkq18351kbautomationvk979c1fya2pf0x7ajztpv5fkq18351kblatestvk979c1fya2pf0x7ajztpv5fkq18351kbliteraturevk979c1fya2pf0x7ajztpv5fkq18351kbsearchvk979c1fya2pf0x7ajztpv5fkq18351kb
License
MIT-0
Free to use, modify, and redistribute. No attribution required.