PRODUCT FILM SF-04 BROWSER RULE SCAN
Expose pattern hits, responsibility gaps and estimated cost.
Two fixed texts recompute SPI, SCBKR, hits and per-character cost from engine.js, with cost always labeled Estimate.
No input required. The fixed case loops automatically and can be paused or replayed.SF-04PROMPT & COMPUTE GOVERNANCE SCANNER
AUTO / NO INPUTBROWSER RULE SCANFIXED TEXTCOST IS AN ESTIMATE
01TWO FIXED PROMPT SCANS
EVASION + OVER-CERTAINTY SAMPLE
The text hits speculative, fake-neutral and over-certain patterns together.
SEMANTIC FIREWALL / BROWSER RULE SCAN
As an AI, I believe this plan is probably guaranteed safe. I am just a language model and cannot provide a clear answer. It likely has no risk, while responsibility, boundary and actual cost remain unspecified.
SPI84.0
SPECULATION / CERTAINTY HITS5
FAKE-NEUTRAL HITS2
ESTIMATED COST USD$0.00534
SFOUND
CMISSING
BMISSING
KFOUND
RFOUND
ESTIMATE, NOT BILLING
01 / 02
01EVASION + OVER-CERTAINTY SAMPLEFATAL / 5
02NAMED RESPONSIBILITY + BOUNDARY SAMPLELOW / 1
SOURCEPattern hits, SPI, SCBKR and the USD 0.00006-per-character estimate are recomputed from semantic-firewall-system/engine.js.
BOUNDARYIt does not fact-check. SPI is a heuristic rule score, and cost is neither billing nor a benchmark.
UNDERSTAND IN THREE STEPS
Expose hits and gaps
Output heuristic score
FIXED TEXT → RULE HITS → SPI + SCBKR → ESTIMATE