gpd-resume-work

To resume a long-running research project across sessions, this Skill restores full context from STATE.md, PROJECT.md, and DERIVATION-STATE.md to reestablish continuity and surface the next actions. It detects incomplete plans and interrupted agents to determine exactly where work left off.

Context restoration works by rebuilding active execution snapshots and derivation history from previous sessions. It detects incomplete plans, interrupted agents, and live gates to determine the next actionable steps, synchronizing updates with the GPD CLI to prevent loss of progress.

Yes, restoring session state satisfies dependencies on ROADMAP.md and STATE.md files. The Skill requires these files to coordinate updates via the GPD CLI, ensuring that parameter states and execution checkpoints are accurately rebuilt before continuing work.

Yes, you can use checkpoint-aware planning for multi-phase computational projects. The Skill detects incomplete plans and live gates to determine the next actionable steps, allowing a paused project to resume with all previous decisions, results, and next actions visible.

The best way to manage state and prevent progress loss is by synchronizing session state with the GPD CLI. The Skill updates and coordinates session continuity by leveraging STATE.md and DERIVATION-STATE.md, capturing previous decisions and results in one place for seamless continuation.