manage-contextlisted

# Manage Context Recovery-phase skill. Keep work stable when context grows long, the codebase is unfamiliar, attention drifts, or the model is stuck. Combines context budget, session hygiene, zoom-out, deep triage, escalate, onboarding, and post-compact re-anchor. ## Iron Rule <EXTREMELY-IMPORTANT> 1. NEVER continue editing when the same bug surfaces from 3+ angles — that is a context loss signal. Zoom out first. 2. NEVER assume a fact from earlier in the session is still valid for files that have been edited since. Re-read before acting. 3. ESCALATE to a stronger model (or a fresh session) at the third failed attempt at the same goal, not after the tenth. In a debug flow this is debug-issue §9: one cross-model consult, one advisor-informed attempt, then the user. 4. ONBOARDING in an unfamiliar repo: detect stack and conventions from real files before writing a single line. </EXTREMELY-IMPORTANT> ## When to use - An observable context signal fires: the context bar turns yellow/red, the harness warns about compaction, or you notice degraded recall (re-reading files you already read, forgetting stated constraints). When a meter exists, ~70% used is the line — but act on the observable signal, not the estimate - You just resumed from a compaction summary (auto-compact or `/compact`) - A usage / quota limit is near — the session dies regardless of context state; handoff, don't trim - The same bug keeps reappearing at a different surface - You forgot a constraint the user sta