01 · Brief the agent
Copy one prompt. Keep the full context.
Paste this into Claude, Codex, Cursor, ChatGPT, or another coding agent. It already includes the domain, the evidence to inspect, the expected output, and the approval boundary. If this is your first run, complete the one-time setup first; your agent can handle those commands too.
02 · Review the run
Know what happens before you paste.
The agent follows this sequence and reports the evidence in plain English. Any step that sends, uploads, or changes something waits for your explicit approval.
- 01
Diagnose
Run
vdr diagnoseto explain why TrueDR is lower than DR. - 02
Read signals
Run
vdr sites:truedr --detailedto read the owner-scoped signal breakdown. - 03
Review spam risk
Run
vdr sites:disavowonly when severe spam-link risk is the issue. - 04
Find partner fix
Run
vdr opportunitieswhen authority or relevance is the issue. - 05
Re-check
Run
vdr trackafter the next trust refresh to confirm whether the signal improved.
03 · Inspect the commands
Nothing hidden behind the prompt.
These are the commands the prompt asks the agent to run. Copy the whole chain or execute it yourself. Read-only analysis runs immediately; consequential actions keep their approval gate.
# 1. Explain why TrueDR is lower than DRvdr diagnose yourwebsite.com # 2. Read the owner-scoped signal breakdownvdr sites:truedr yourwebsite.com --detailed # 3. Only for severe spam risk, create Google disavow candidatesvdr sites:disavow yourwebsite.com --min-spam 50 --limit 100 # 4. If authority/relevance is the issue, find one partner fixvdr opportunities yourwebsite.com # 5. Re-check after the next trust refreshvdr track yourwebsite.comNext job
