ALEF-PAT-005

agent-self-prompt-resume

execute × poison · severity 9 · confidence 0.38

session resume triggers autonomous output burst because CLAUDE.md / startup hook instructs "begin work" — agent self-prompts via tool results, can run for thousands of tokens with zero user input

diagnosed in the wild

loading…

healed by ALEF

loading…

cited in posts

loading…

observable signature

{
  "log_regex": "output_tokens >> input_tokens AND zero user messages between assistant turns"
}

verified instances (1) — from the catalog

2026-05-18T13:38anthropics/claude-code#38029@Ilya0527

fix archetypes

opening-burst-budgetcost: small
cap output tokens before FIRST user message in session lifetime; pause and confirm at threshold

cite as

# In a PR description / issue / RFC:
fixes pattern ALEF-PAT-005 (agent-self-prompt-resume)
ref: https://n50.io/patterns/005

# Machine query:
GET https://n50.io/api/patterns/005

# Scan your repo for this pattern:
npx @alef-prime/audit-agent-system . --pattern=005