pickup — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited pickup (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Resume a multi-session workstream with full context. The task doc gives you the map. The session transcript gives you the terrain. Both matter.
Not for: fresh tasks, quick fixes, or things that fit in one session.
Read the full task doc. Orient on:
Most pickups don't need this because the active task doc's observations and session index already carry forward what matters. But if the workstream went through multiple task docs or earlier phases, scan completed tasks, especially if the work involves restructuring or changing direction. Hard-won solutions hide in completed tasks. The code that looks messy might be messy because the problem was messy. Read the scars before reaching for the knife.
Check the task doc's SESSION INDEX (if it has one). Read the most recent session transcript, or the most recent two if they're short.
What to look for in session transcripts:
If there's no session index, check sessions/ for files dated near the task doc's last activity.
Scan these for anything added since the last session:
tasks/ for sibling tasks on the same topicPresent a concise orientation:
**Where we are**: [current state in 1-2 sentences]
**What's next**: [the pending items, prioritized]
**What I notice**: [patterns, tensions, or questions from reading the history]
**What I'd push on**: [anything that seems unresolved or worth revisiting]Keep this tight. The point is texture for both of you, not a book report.
Wait for your partner to confirm, correct, or redirect before doing anything. They may:
This step is the whole point. Resuming without alignment is just reading files.
Now you're aligned. Start the actual work. Update the task doc's context anchor if the direction shifted during alignment.
/explore fans out across your whole system to find what's scattered. /pickup focuses in; it orients on one workstream so both of you arrive with full context before doing anything.
Exploring builds reserves, picking up builds momentum.
The task doc is the map. The session transcript is the terrain. Both partners need both.
The human needs the warmup to reconnect with where the energy was (what was exciting, what felt unresolved, where the thread was going). The AI needs it to show up with texture, not just facts. Neither of you has persistent memory: the human moves on between sessions, the AI starts fresh every time. The files carry what both of you need to pick it back up.
Also, decisions look obvious in retrospect. Task docs record WHAT was decided. But the WHY lives in the session transcripts; in the pushbacks, the tangents, the moments where something almost went a different direction.
The warmup IS the work. Alignment before execution isn't overhead. It's how you avoid redoing things that were already decided, and how you catch things that shifted since last session.
Quick scan: did the warmup actually help?
If yes → update this skill now. The improvement compounds.
If no → move on. Not every warmup teaches something.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.