ワンクリックで
bip-epic-tuckin
Persist orchestrator state before context reset
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
メニュー
Persist orchestrator state before context reset
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Check remote server CPU, memory, and GPU availability via SSH
Cold-start into a worktree/clone from a fresh conversation — read the PR, issue, and any status files, figure out where things stand, then STOP and ask the user what to do next. Use for a fresh conversation dropped into a bip-spawn or bip-epic-spawn worktree/clone that already has history (a PR, an in-progress phase, or a stalled worker).
Quick poll of tracked EPICs and code repos for new manuscript-relevant results
Persist manuscript session state before context reset
Cold-start for a manuscript session — the paper is the source of truth and shared context; discuss results, orchestrate research through issues/PRs, and update the paper as threads complete
Spawn a Claude session in a clone for an EPIC issue
| name | bip-epic-tuckin |
| description | Persist orchestrator state before context reset |
Flush orchestrator state to durable storage before a context reset or session end. Run this when context is getting long or before stopping.
/bip-epic-tuckin
For each ISSUE-EPIC-*.md file in the repo root, follow the EPIC body
update pattern from /bip-epic (pull with updatedAt tracking → edit →
conflict-check → push). Extract the issue number from the filename
(ISSUE-EPIC-284.md → 284).
Report which EPICs were pushed (and which were skipped due to conflicts).
Read clone_root and local_worktrees from .epic-config.json.
Slot paths:
$CLONE_ROOT/<clone-name>$CLONE_ROOT/issue-<N>For each slot the orchestrator interacted with this session:
.epic-status.json is stale or missingCLONE_ROOT=$(jq -r .clone_root .epic-config.json)
cat > "$CLONE_ROOT/<slot>/.epic-status.json" << 'EOF'
{
"issue": <N>,
"title": "<title>",
"phase": "<current phase>",
"summary": "<what happened>",
"updated_at": "<ISO 8601 timestamp>",
"blockers": []
}
EOF
Update slots the orchestrator has direct knowledge about. This includes:
Do not guess status for slots with active sessions that may have progressed beyond what the orchestrator last observed.
Most state should already be in EPIC bodies (ISSUE-EPIC-<N>.md) and
clone status files (.epic-status.json). Only update MEMORY.md for
orchestrator-level context that doesn't fit in those files:
Print a summary:
## Tuckin Complete
- EPICs pushed: i281, i295
- EPICs skipped (conflict): i310
- Slot status updated: cedar, issue-295
- MEMORY.md updated: yes
Safe to reset context.