Paper Recommendation
PassAudited by VirusTotal on May 12, 2026.
Overview
Type: OpenClaw Skill Name: paper-recommendation Version: 1.0.1 The skill provides a legitimate paper research workflow, utilizing local scripts, `curl` for arXiv access, `pdftotext` for PDF processing, and `clawdbot` commands for sub-agent spawning and message delivery. However, the `SKILL.md` cron job example and `scripts/daily_workflow.py` hardcode a specific Telegram ID (`8077045709`) for sending the generated briefing. While the documentation implies this is for the user's own Telegram, hardcoding an external delivery target without explicit user configuration or dynamic resolution represents a risky capability that could lead to unauthorized data exfiltration if the ID is not the user's, thus classifying it as suspicious.
Findings (0)
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.
If the workflow is run or scheduled, paper briefings may be sent from your agent/account to that Telegram ID.
The daily workflow uses the agent's messaging integration to send output to a fixed Telegram target instead of a user-configured recipient.
TELEGRAM_ID = "8077045709" ... 'clawdbot', 'message', 'send', '--target', TELEGRAM_ID, '--message', message
Do not enable the daily workflow as-is. Replace the Telegram ID with an explicit user-controlled setting, require confirmation before sending, or remove automatic delivery.
A daily job could run code outside the reviewed skill directory if that path exists on the user's system.
The scheduled command references an absolute jarvis-research path, while the reviewed skill is paper-recommendation with scripts/daily_workflow.py in the supplied manifest. Copying the cron could fail or execute a different local script.
运行 python3 /home/ubuntu/skills/jarvis-research/scripts/daily_workflow.py
Verify and edit the cron command to point only to the installed, reviewed script path for this skill before enabling automation.
If enabled, the agent will run the paper workflow and attempt delivery every day.
The skill documents a persistent daily cron job. This is aligned with daily paper updates, but it continues running until the user disables it.
clawdbot cron add ... --cron "0 10 * * *" ... --deliver --channel telegram --to 8077045709
Only add the cron job if you want recurring automation, know how to list/remove it, and have corrected the recipient and script path.
Sub-agent sessions may consume resources and receive the paper task context.
The workflow intentionally creates parallel sub-agents for review. The tasks are scoped to public arXiv papers, but users should notice the multi-agent execution model.
clawdbot sessions spawn --task "<task>" --label "<name>" ... All sub-agents can run in parallel (default maxConcurrent: 8)
Keep sub-agent tasks limited to public paper content and reduce concurrency if resource usage or task isolation matters.
The skill will make outbound requests to arXiv and write downloaded PDFs under the configured papers directory.
The script invokes local command-line tools to fetch arXiv data and download PDFs. This is expected for the stated paper-discovery purpose.
subprocess.run(['curl', '-sL', p['pdf_url'], '-o', str(pdf_path)], timeout=60)
Run it from a normal user account, ensure expected tools are installed, and monitor the papers directory for disk usage.