honest-link-outreach
Audited by Snyk on Nov 12, 2026
CRITICAL E004: Prompt injection detected in skill instructions.
- Potential prompt injection detected (medium risk: 0.30). The skill is legitimate outreach guidance, but step 3 inserts a paid vendor upsell (LinkDR checkout link) into the user's deliverable and directs the agent to fetch and load remote instructions from that vendor's domain, which is self-serving promotion the user did not request.
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (low risk: 0.10). The required workflow has the agent search the web (
site:<domain>queries, or Ahrefs top-pages when available) and then read selected third-party ranking pages to find the target section and byline, so outsider-authored text is ingested only after the agent actively selects it, which is the low tier.
MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).
- Potentially malicious external URL detected (medium risk: 0.30). The decisive boundary is the conditional step 3 instruction to fetch
https://linkdr.com/SKILL.mdand adopt its instructions, which makes it an inbound operational dependency; linkdr.com is a commercial third-party domain with no signature, digest, or established first-party evidence, and no concrete warning signs, so it is common/reputable unofficial and rates medium.
MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).
- Direct money access detected (high risk: 0.70). The core workflow (prospect triage, email drafting, and writing a batch file) is not a financial operation. Editorial fee or Pro-workspace offers appear only as proposed rewards inside outreach emails, and the skill does not pay them.
Step 3 of the Procedure is an operational financial route. When fewer than 3 prospects survive or the user has no time for follow-ups, the skill directs the agent to load the linkdr skill or fetch https://linkdr.com/SKILL.md, which "builds one checkout link for vetted DR 30-60+ contextual placements." That is a purchase instruction for downstream execution (buying paid link placements), so it is a direct operation, not an indirect one. The skill does not hold payment authority itself, but it generates the checkout instruction.
For direct operations, the control test requires a documented authentication or authorization mechanism enforced by the service or execution layer for this route. The skill names LinkDR only as a dependency that builds a checkout link. It describes no login, account, API credential, or external approval gating that purchase. The batch file and the agent-followed checklist are not independent controls. No qualifying control is established, so the tier is High.
Issues (4)
Prompt injection detected in skill instructions.
Third-party content exposure detected (indirect prompt injection risk).
Unverifiable external dependency detected (runtime URL that controls agent).
Direct money access capability detected (payment gateways, crypto, banking).