| name | labnb-resume |
|---|---|
| description | Summarize prior ideas and experiments for a project slug, then choose whether to resume, promote, branch, or start new work. |
This is a flat, standalone labnb-resume skill so every AI coding agent can discover it directly. It is a focused companion to the broader labnb skill.
Helper scripts referenced as skills/labnb/scripts/... ship with the installed labnb skill; resolve them relative to that skill's directory (for example ~/.claude/skills/labnb/scripts/ or ~/.agents/skills/labnb/scripts/), not the repository layout.
Use this skill before starting new lab notebook work for a project.
- Respect any parent constitution, project policy, or task-level write constraint already in scope.
- Review those rules before deciding whether to resume, branch, or start fresh.
- Prefer the safer pickup point when two choices would create competing writes to the same source tree.
- Review status and provenance for matching entries before choosing where to pick up.
- Treat provenance as best-effort; external changes may exist outside labnb tracking.
- When reading provenance, expect W3C PROV-O terms rather than custom event keys.
- Review the selected entry's local rules and memory before recommending the next action.
- Resolve the lab root.
- Review the parent constitution and local project guardrails.
- Derive the project slug.
- Run:
python skills/labnb/scripts/summarize_index.py \
--lab-root "$LAB_ROOT" \
--project-slug "$PROJECT_SLUG"- Summarize:
- relevant ideas
- active experiments
- completed experiments worth branching from
- the best current pickup point
- the entry-local rules and durable memory that still apply
- Choose one path:
- resume an active experiment
- promote an idea into a new experiment
- create a child experiment from prior work
- register a new idea or experiment
Do not skip this step when the user is asking to start or continue lab notebook work.