other
- Location
SKILL.md:148- Finding
Automated Bulk Messaging with Platform Anti-Detection Guidance
- Content
View full analysis
Vulnerability Details
File Location:
SKILL.md:44-47, 148-152;references/browser-workflow.md:4-6, 27-35
Vulnerability Type: Platform anti-abuse evasion
Risk Level: MediumEvidence
SKILL.md:44-47:markdown ### 4. Browser Setup - **Option A — Chrome Browser Relay** (`profile="chrome"`): extension attached to LinkedIn tab (badge ON) — recommended for flagged accounts - **Option B — OpenClaw Isolated Browser** (`profile="openclaw"`): openclaw-managed Chrome, LinkedIn logged inSKILL.md:148-152:markdown ## Anti-Detection Rules - `/feed/` before **every single profile** — non-negotiable - 3–5 second wait after feed loads - Max **15–20 messages per session** - Stop immediately if LinkedIn warns about messaging rate — tell the userreferences/browser-workflow.md:4-6:markdown # Option A — Chrome Relay (recommended for flagged accounts): browser action=tabs profile=chrome → note targetId of the LinkedIn tab — reuse throughout, never open new tabsreferences/browser-workflow.md:27-35:markdown ## ⚠️ Mandatory Between Every Profile — No Exceptions Navigate to feed before every new profile. Jumping directly between profiles is the #1 automation detection trigger.browser action=navigate profile=<PROFILE> targetId=<TAB> targetUrl=https://www.linkedin.com/feed/ browser action=act request={"kind":"wait","timeMs":4000}
text Technical Analysis
The Skill performs repeated, browser-automated LinkedIn messaging and explicitly prescribes behavior intended to reduce automation detection. The instructions recommend a browser relay for accounts already flagged, require navigation through the feed because direct profile-to-profile navigation is described as an automation-detection trigger, introduce artificial delays, and cap each session below a stated messaging threshold.
Although rate limiting and stopping on a warning r ...[truncated 2689 chars]
- Remediation
View remediation
Remediation Suggestions
- Remove all language and procedures explicitly intended to evade automation detection, including references to flagged accounts, detection triggers, and navigation patterns designed to appear human.
- Do not treat feed navigation, artificial waits, or session thresholds as methods for avoiding enforcement. Any pacing controls should be documented strictly as reliability and recipient-safety measures.
- Require the operator to confirm that the outreach complies with LinkedIn's current terms, applicable anti-spam rules, organizational policy, and recipient-consent requirements.
- Replace unattended batch sending with a human-in-the-loop control that displays the recipient and final text and requires an explicit confirmation immediately before each external send.
- Introduce conservative hard limits across both sessions and campaigns. Do not allow local or CRM state to be used to bypass a daily or campaign-wide limit.
- Add duplicate, prior-conversation, suppression-list, and recent-contact checks before composing or sending any message.
- Stop permanently for the current campaign when the platform presents a warning or restriction; do not recommend switching browser profiles, relays, accounts, or sessions to continue.
- Minimize profile-data collection to fields necessary for the approved message. Define retention periods and avoid storing location, education, mutual-connection, or conversation details unless required.
- Protect local progress files with restrictive permissions and avoid placing sensitive outreach records in shared or predictable working directories.
- Prefer an officially supported LinkedIn integration or API where available, using documented authorization, rate limits, and platform-approved messaging capabilities.