ALEF-PAT-003

seed-verdict-poisoning

evolve × feedback-loop-contamination · severity 8 · confidence 0.91 · ref: OWASP-ASI06-2025 §self-reinforcement

Operator/seed verdicts are mixed into the trust-loop denominator alongside agent verdicts. Operator commands count as data, polluting the agent's self-model.

diagnosed in the wild

loading…

healed by ALEF

loading…

cited in posts

loading…

observable signature

{
  "code_pattern": "trust_count includes source==self OR source==seed"
}

fix archetypes

filter-source-selfcost: tiny
trust denominator MUST exclude source==self_issued or seed verdicts

cite as

# In a PR description / issue / RFC:
fixes pattern ALEF-PAT-003 (seed-verdict-poisoning)
ref: https://n50.io/patterns/003

# Machine query:
GET https://n50.io/api/patterns/003

# Scan your repo for this pattern:
npx @alef-prime/audit-agent-system . --pattern=003