esper-continue

You are resuming an interrupted implementation session.

Step 1: Check setup

Run esperkit context get. If it fails, tell the user to run esper:init first and stop.

Parse the context JSON.

Step 2: Find the active increment

If active_increment is null:

• Run esperkit increment list --format json to check for pending increments
• If pending increments exist: "No active increment. You have pending increments. Run esper:atom to activate one, or esper:batch to review the queue."
• If no increments at all: "No increments found. Run esper:atom to create one, or esper:spec to work on specs."
• Stop.

Step 3: Read the active increment

Read the increment file from .esper/increments/active/<filename>. Parse its frontmatter and full body. If the increment references a spec file, read it: esperkit spec get <file>

Step 4: Assess current state

Run git log --oneline -10 to see recent commits. Run git status --porcelain to see uncommitted changes.

Cross-reference against the increment:

• Check ## Scope items against recent commits and file changes
• Check ## Progress section for recorded progress
• Check ## Files Affected against actual file state
• Check ## Verification — run the verification commands if possible

Step 5: Summarize and resume

Present a concise status:

Resuming: [increment title]

Done:
  - [completed scope items]

Remaining:
  - [incomplete scope items]

Uncommitted changes:
  - [files with changes]

Step 6: Continue implementation

Pick up implementation from the first incomplete scope item:

1. Follow the ## Scope section
2. Create/modify files as planned
3. Run verification after each change
4. Update ## Progress as work is completed
5. Commit according to workflow_defaults

When all scope items are complete:

• Run final verification
• Ask the user if they want to finish: "All scope items complete. Run esper:go to finish, or esper:review to verify first."

Available CLI commands

• esperkit context get — read context
• esperkit increment list — list increments
• esperkit spec get <file> — read a spec