Post-Leak Claude Code Playbook
- Entity ID:
ent-20260423-r30a00000001a - Type:
concept - Scope:
private - Status:
active
Description
De facto seven-step operational pattern emerging in community posts since April 15: (1) run via CLI + native binary and avoid repacked binaries that break attestation; (2) design your own system prompt layer; (3) optimize for caching and context (treat SYSTEM_PROMPT_DYNAMIC_BOUNDARY as sacred); (4) pick effort/model deliberately (reserve xhigh for truly hard tasks); (5) lock down sandbox.fs and sandbox.network; (6) exploit desktop multi-agent capabilities (side chats, parallel sessions); (7) instrument your own usage with /insights and custom JSONL scanners.
Key claims
- Post-leak playbook: seven operational adaptations
- Accept v2.1.113 native binary; avoid repacked binaries
- Use Opus 4.7 high for most coding; reserve xhigh for hard reasoning
Relations
- Post-Leak Claude Code Playbook --[depends_on]--> Native Claude Code Binary Spawn
- Post-Leak Claude Code Playbook --[depends_on]--> sandbox.network.deniedDomains Setting
- Post-Leak Claude Code Playbook --[depends_on]--> xhigh Effort Level