gal-tab/sir_albert/resume
- community
Checking that the SKILL.md file is still in the repository.
Reload the latest handoff for the current project and state the plan back concisely before doing anything, so the user gets productive in ≤2 turns with zero re-explaining. Use when the user says "resume", "/resume", "pick up where we left off", "load the handoff", "continue last session", "what were we doing", or "reload context".
Install
npx vskill@latest install gal-tab/sir_albert/resumeSKILL.md
Resume
Reload the latest handoff and state the plan back in ≤5 bullets — BEFORE taking any other action.
Announce at start: "I'm using the resume skill to reload your session."
1. Boot from the OS preamble
Load os/PREAMBLE.md (USER identity + active rules + anti-slop / use-the-skill contract) before proceeding. This grounds the resumed session in who the user is and what rules apply.
2. Find the latest handoff
Look for .memory-bank/HANDOFF*.md in the current project directory.
- Sort candidates by filename date (HANDOFF-YYYY-MM-DD.md) or mtime — take
the most recent.
- If
$ARGUMENTSnames a specific handoff or date, prefer that file. - If no handoff exists: say so plainly, then fall back:
- Run
git log --oneline -10andgit branch --show-current. - Check for any
.memory-bank/*BUILD*.mdor.memory-bank/*PLAN*.mdledger. - Summarise what you find, then ask: "No handoff found — what do you want to
focus on?" Do not proceed past this point until there is something concrete to resume.
3. Read it fully + confirm working state
- Read the handoff doc end-to-end.
- Run:
- Compare the current branch and dirty-file list to what the handoff recorded.
- Mismatch (wrong branch, unexpected dirty files, staged changes not in
handoff): flag it explicitly before the plan bullets. Example: > "Warning: handoff was on feat/xyz but you are currently on main."
- Clean match: proceed silently.
4. State the plan back — ≤5 bullets, TLDR first
Before doing anything else, emit a tight recap. No slop, no filler.
Format:
Cap at 5 bullets total. If there is nothing in a slot, omit that line.
5. Proceed — but respect ask-before-live
After the plan bullets are confirmed (or the user says "go"), begin the next task.
- Do NOT auto-run live or irreversible actions (deploys, migrations, pushes,
destructive file ops) without confirming first, per os/rules/autonomy.md.
- If the next task is ambiguous, ask one focused clarifying question before starting.
Style rules
- TLDR first. The user opened a fresh session — orient them in ≤5 seconds.
- Prescriptive, not descriptive. Bullets are actions or facts, not narrative.
- Reference, don't duplicate. For long docs, reference by path — do not paste
their content into the recap.
- No slop. No "as we discussed", "it's worth noting", or similar filler.
Every bullet earns its place.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingplugins/sir-albert/skills/core/resume/SKILL.md
Popularity
- 7-day trend score
- 0
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/gal-tab/sir_albert/resume)