There's a better way
The Old Way
- One AI call judges relevance, quality, and author together
- A keyword with two meanings drags in the wrong conversation entirely
- When a pick is wrong there is no way to tell which judgment failed
- Every topic shares one threshold whether it suits them or not
The BlackOps Way
- Three independent checks, each with its own score and reason
- Relevance is judged against the brief, never the search term
- Every verdict is saved, so you can see why a tweet was picked
- Score weights tune per hunt
Why Hunt Missions?
Being seen in the wrong thread costs more than missing the right one
A brief, not just keywords
A hunt carries a target brief describing what you are actually looking for. Keywords surface candidates; the brief decides whether they count.
Three scores, not one opinion
Relevance, slop, and author fit run as separate checks in parallel. When a pick is wrong you can see which check failed instead of guessing.
Slop is a hard veto
Engagement farming, recycled demos posted as news, follow-to-get-the-guide mechanics. On-topic does not save it. Your name goes under that reply.
Author fit matters
Practitioners talking about their own work score high. Aggregators, brand accounts, and engagement farms score low. Having a product in your bio is a positive signal, not a strike.
How It Works
Define the hunt
Name it, write the target brief, add keywords, set author allow and deny lists, and bind the brains that give it context.
Candidates get scored
The relevance check never sees the keyword that surfaced the tweet, on purpose. Telling it the search term invites it to confirm the match, which is exactly how homonym false positives get through.
Tune the blend
Scores combine into a weighted composite and fixed rules derive the action. Weights are configurable per hunt, so a technical hunt and a broad one do not have to share a threshold.