/open

Orient to previous work, then act on what's next.

Structure

Gather  → script + optional beads gate
Orient  → "here's where we left off"
Decide  → user picks direction
Act     → draw-down to TodoWrite

1. Gather

Run the context gathering script (from the session-management skill directory):

~/.claude/skills/session-management/scripts/open-context.sh

Script outputs: HANDOFF, GIT, BEADS (if present), UPDATE_NEWS context.

Gate check: If GATE_REQUIRED=true and beads skill is available:

Skill(beads)

Do not proceed until loaded. This ensures draw-down patterns are available.

2. Orient

Synthesize what matters from script output:

• Lead with handoff — Done, Next, any Gotchas
• Surface commands — if Commands section exists, offer to run them
• Note scope mismatches — if handoff "Next" doesn't match bd ready, flag it
• Beads ready — what's unblocked

Keep it brief. Orientation, not full briefing.

3. Decide

User picks direction:

• A specific bead from ready list
• Continue from handoff's "Next"
• Something else entirely

4. Act

Draw-down is mandatory. When user picks a bead:

1. bd show <bead-id> --json — read design and acceptance criteria
2. Create TodoWrite items from acceptance criteria
3. Show user: "Breaking this down into: [list]. Sound right?"
4. Mark bead in_progress: bd update <id> --status in_progress

No TodoWrite items = No work. This creates checkpoints where drift gets caught.

Graceful Degradation

Not all features require all dependencies:

The core pattern (Gather → Orient → Decide → Act) works with just git status.

Mirrors