Truncate Aethervault sessions to reclaim token space during context overflow.
npx skills add https://github.com/sunilkgrao/aethervault --skill emergency-compact
Please help me install this Agent Skill. Skill: emergency-compact Source: https://github.com/sunilkgrao/aethervault/tree/main/skills/emergency-compact Command: npx skills add https://github.com/sunilkgrao/aethervault --skill emergency-compact
Emergency Compact forces context reduction by truncating older messages when normal compaction fails or context overflow persists, helping conversations continue smoothly without data loss.
Run the emergency-compact script to truncate the current session when context overflows.
High-intent search queries and answers about installing and using this skill.
Emergency context compaction forcibly truncates older messages to reclaim available token space when normal auto-compaction fails. It backs up the current session, truncates to the last N messages, and appends a system truncation notice.
Context truncation reduces active context size by cutting off older messages and keeping only the last N messages. It reclaims token space by appending a system truncation notice, allowing long-running sessions to continue without data loss.
You can handle repeated 413 prompts by applying an aggressive fallback compaction. This forcibly truncates the active session to the last N messages, reducing memory usage and stopping persistent context overflow errors.
Yes, emergency compact creates a truncated backup of the current session before removing older messages. It also logs a system truncation notice, ensuring you retain a record of the session before reclaiming token space.
You should use an aggressive fallback for session management when normal auto-compaction fails or context overflow persists. It is designed for stubborn sessions with high context usage that repeatedly trigger 413 prompts.
KodaTCGMonitors and compresses AI session tokens to prevent 413 context overflow errors.
Kgan01Detect context degradation and apply recovery strategies in long-running tasks.
LincyawSuggest manual context compaction after a configurable threshold of tool calls.
SJTU-IPADSSearch and analyze JSONL session logs with jq to filter messages by keyword and timestamp.
