AC screening-alert-adjudication-amir-fadavi
Adjudicates whether a hit generated by sanctions, PEP, or adverse-media screening is a true positive, false positive, or requires human escalation. Use whenever a user presents a screening alert, a name match against a watchlist (OFAC SDN, EU consolidated list, UK OFSI, UN list, PEP list, adverse media hit, etc.), or asks to clear a screening hit / reduce false positives / determine whether a flagged name is actually the listed party. Use even when the user describes the task casually — "is this person actually on the sanctions list", "did we get a real match", "clear this alert", "I have a hit on X" — these are all screening-adjudication tasks. Produces a deterministic determination with full audit trail (structured JSON + human-readable narrative). Designed for use by compliance analysts and screening systems.
Adjudicates whether a hit generated by sanctions, PEP, or adverse-media screening is a true positive, false positive, or requires human escalation.
As a process C 64/100 · Has gaps — weak spots: inputs and preconditions, failures and branches
How to improve
- A spec.yaml with trigger phrases and assertions — a behaviour contract for CI; `skilltest init` writes a template.
Guard findings · 0
✓ No critical or high findings
Files scanned: 11. Evidence is masked. Grey chips explain why severity was lowered.
Against the Agent Skills spec
✓ No remarks against the Agent Skills spec
Process rating: all ten parameters 64/100
- 0Inputs and preconditions. Does not say what the process needs to start
- 0Failures and branches. Linear process with no failure handling
- 60Tools and files. Uses tools (web) that frontmatter does not declare
- 60Result and completion. Output format stated, no completion criterion
- 85Steps. 44 steps, 1 vague phrases
- 100When it triggers. States when to use and when not to
- 100Consistency. Name and required fields are in place
- 100Execution cost. Instruction body is 3069 tokens
- 100Running it twice. No mutating operations
- 100Progress reporting. Reports progress
- medium Safety rules and hard prohibitions inside a skill: they belong in the system prompt, here they protect nothing
- low 11 top-level sections: this looks like several domains in one skill
- medium 5 test cases, all positive: not one "should refuse" or "should ask first"
- low No test case covers injection arriving through data
Everything here is measured from the skill text rather than judged by a model, so the numbers are checkable. A parameter weighs more when it is a more common reason for the process to stall.
Quality signals
- +4Description does not say when NOT to use the skill (false activations)
- +3Description length 823: 120–800 characters recommended
- +4No input/output examples
- +2Single-language instructions
- +5Description quotes 4 example trigger phrases
- +4Structure: 19 headings
- +3Step-by-step instructions: 44 items
- +3Output format is stated explicitly
- +4Reference files are cited in the instructions (8 of 8)
- +1License stated
Quality base 70; lint remarks subtract, signals add up to 100. Result: 90.