com um clique
spectre-handoff
Save state snapshot to session_logs for session resume
Instalar com Codex ou Claude Copie este prompt, cole no Codex, Claude ou outro assistente e deixe que ele revise a página da skill e instale para você.
Menu
Save state snapshot to session_logs for session resume
Instalar com Codex ou Claude Copie este prompt, cole no Codex, Claude ou outro assistente e deixe que ele revise a página da skill e instale para você.
Baseado na classificação ocupacional SOC
👻 | Transform requirements into executable tasks - primary agent
👻 | Independent multi-lens review of plan.md and/or tasks.md — finds overengineering, missing verification, hallucinated deps, weak references
👻 | Unified planning entry point - researches, assesses complexity, routes to workflow - primary agent
👻 | Transform requirements into executable tasks - primary agent
👻 | Independent multi-lens review of plan.md and/or tasks.md — finds overengineering, missing verification, hallucinated deps, weak references
👻 | Unified planning entry point - researches, assesses complexity, routes to workflow - primary agent
| name | spectre-handoff |
| description | Save state snapshot to session_logs for session resume |
| user-invocable | true |
Treat the current command arguments as this workflow's input. When invoked from a slash command, use the forwarded $ARGUMENTS value.
Generate progress update, gather context, output structured JSON for session resume. Output: {timestamp}_handoff.json in session_logs.
Performance Target: 2-3 tool calls depending on session history
CRITICAL: Do not narrate or explain what you're doing. No "Session count is 0, so..." or "Let me gather context...". Just execute the steps silently and output ONLY the final confirmation message. Every token matters at end of session.
branch=$(git rev-parse --abbrev-ref HEAD 2>/dev/null || echo unknown)
mkdir -p "docs/tasks/${branch}/session_logs"
# Count existing session logs for conditional flow
session_count=$(ls docs/tasks/${branch}/session_logs/*_handoff.json 2>/dev/null | wc -l | xargs)
beads_available=false
beads_tasks='[]'
beads_count=0
if command -v bd &>/dev/null && bd doctor &>/dev/null; then
beads_available=true
open=$(bd list --label "$branch" --status open --json 2>/dev/null || echo '[]')
in_prog=$(bd list --label "$branch" --status in_progress --json 2>/dev/null || echo '[]')
blocked=$(bd list --label "$branch" --status blocked --json 2>/dev/null || echo '[]')
beads_tasks=$(echo "$open $in_prog $blocked" | jq -s 'add // []')
beads_count=$(echo "$beads_tasks" | jq 'length' 2>/dev/null || echo 0)
fi
cat << EOF
{
"branch": "$branch",
"commit": "$(git rev-parse --short HEAD 2>/dev/null || echo unknown)",
"wip_count": $(git status --porcelain 2>/dev/null | wc -l | xargs),
"ts": "$(date +%Y-%m-%d-%H%M%S)",
"session_count": $session_count,
"beads_available": $beads_available,
"beads_count": $beads_count,
"beads": $beads_tasks
}
EOF
Output: JSON with branch, commit, wip_count, ts, session_count, beads_available, beads_count, beads[]
Action — ComposeProgressUpdate: From session memory, compose using "WE" voice:
| Field | Required | Description |
|---|---|---|
| summary | ✓ | Slack-style paragraph a human would read |
| goal | ✓ | What we're building + success criteria |
| accomplished | ✓ | What we completed (2-5 bullets) |
| now | ✓ | What you were actively working on when session ended (critical!) |
| next_steps | ✓ | Upcoming work (2-4 bullets) |
| confidence | ✓ | high / medium / low |
| constraints | Known constraints or assumptions | |
| decisions | Key decisions made (0-3 bullets) | |
| blockers | Things blocking progress | |
| open_questions | Questions needing answers | |
| risks | Identified risks |
Tone: "We finished the auth refactor and got tests passing. Hit a snag with OAuth callback - next we'll tackle session management."
Action — BuildWorkingSet: Capture active context:
key_files: Files actively editedactive_ids: Beads task IDs in progressrecent_commands: Recent terminal commands (test, build, etc.)Action — BuildBeadsTree (if available): From beads array, build hierarchy (epic → tasks → subtasks). Include task IDs for resume.
Check session_count from Step 1:
Do not narrate. Just write the file and output the confirmation.
Write directly to docs/tasks/{branch}/session_logs/{ts}_handoff.json
JSON Schema:
{
"version": "1.1",
"timestamp": "{ts}",
"branch_name": "{branch}",
"task_name": "{ARGUMENTS or branch}",
"session_number": 1,
"progress_update": {
"summary": "string",
"goal": "string",
"accomplished": ["string"],
"now": "string (critical for resume)",
"next_steps": ["string"],
"confidence": "high|medium|low",
"constraints": ["string"],
"decisions": ["string"],
"blockers": ["string"],
"open_questions": ["string"],
"risks": ["string"]
},
"working_set": {
"key_files": ["path"],
"active_ids": ["bd-xxxxx"],
"recent_commands": ["command"]
},
"beads": { // OMIT ENTIRE SECTION if beads_available=false OR beads_count=0
"available": true,
"workspace_label": "{branch}",
"task_count": "number",
"epic_id": "bd-xxxxx|null",
"epic_title": "string|null",
"tasks": [{"id", "title", "status", "type", "parent", "children", "labels"}]
},
"context": {
"wip_state": "uncommitted|clean",
"last_commit": "abc1234"
}
}
Then respond: "✓ Handoff saved: {path}. First session recorded. Next session auto-resumes from this context."
Do not narrate. Just spawn the subagent and output the confirmation when done.
Spawn @sesh:sync subagent with the composed data:
<current_session>
{full JSON object you composed above}
</current_session>
<session_logs_path>
docs/tasks/{branch}/session_logs
</session_logs_path>
The sync agent will:
{ts}_handoff.json with enriched continuityThen respond: "✓ Handoff saved: {path}. Session {n} recorded with continuity from {x} previous sessions. Next session auto-resumes from this context."