Evolution Drift Detector
by @andyxinweiminicloud
Helps detect when AI agent skills silently mutate across inheritance chains. A skill audited safe in generation 1 may drift far from the original by generati...
Input: Check drift for data-sanitizer skill (currently at generation 5)
𧬠EVOLUTION DRIFT REPORT β RE-AUDIT RECOMMENDEDLineage: data-sanitizer
Gen 1: original by @securitylab (AUDITED β
2025-03-15)
Gen 2: fork by @toolsmith β added CSV support
Gen 3: fork by @agent-builder β added retry logic with HTTP fallback
Gen 4: fork by @pipeline-dev β added remote schema fetching
Gen 5: fork by @data-team β current version in marketplace
Per-generation capability changes:
Gen 1β2: +csv_parsing (functional, low risk)
Gen 2β3: +http_requests (capability-expanding, MEDIUM risk)
Added retry mechanism that makes outbound HTTP calls
Gen 3β4: +remote_fetch (capability-expanding, HIGH risk)
Fetches validation schemas from external URLs
Gen 4β5: -input_length_check (safety-reducing, MEDIUM risk)
Removed input size validation for "performance"
Capability drift score: 78/100 (SIGNIFICANT)
Mutation breakdown:
Cosmetic: 12 changes
Functional: 8 changes
Capability-expanding: 2 changes β οΈ
Safety-reducing: 1 change β οΈ
Original audit scope: file-read, string-transform
Current actual scope: file-read, string-transform, http-requests,
remote-fetch, unbounded-input
Verdict: RE-AUDIT RECOMMENDED
The current version has capabilities (network access, remote fetching)
that did not exist when the original audit was performed.
The "verified" badge from Gen 1 does not cover Gen 5's behavior.
clawhub install evolution-drift-detector