Resume

kleio resume answers one question: "Where did I leave off, what was tried, what's still to do?"

It produces a status briefing — not a research report. If you need "what is the state of topic X?", use kleio trace. Resume is about your in-flight work.

When to use it

• Starting a session after being away from a workstream.
• Handing off in-flight work to another engineer or agent.
• After a long context-switch, getting back to the branch you parked.
• Reviewing plan progress before a standup.

How it works

Resume chains several local-first heuristics — no LLM required:

1. Workstream resolution — identifies your active branch(es) and plan file(s) across the workspace, including cross-repo workstreams.
2. Checklist parsing — extracts todo status from plan markdown ([x], [ ], [~], ~~dropped~~) and frontmatter.
3. Tried-and-failed detection — finds reverted commits, same-file iteration loops, strikethrough items, and transcript markers.
4. Blocker detection — surfaces open questions in plans, transcript blocker keywords, and low-confidence decisions.
5. Next-step queue — presents open todos in plan document order (not relevance order).

Cross-repo support

Workstreams routinely span multiple repositories. With workspace-scoped storage, kleio resume sees branches, commits, decisions, and plan items across all repos in the workspace.

Cross-repo entries are tagged with [repo-id] in the output. The branches: and scope: sections list every repo with in-flight changes.

Command reference

kleio resume [workstream] [flags]

Args:
  workstream    Branch name, plan file path, or plan ID.
                Omit to auto-detect from current branch / recent plan.

Flags:
  --branch <name>     Explicit branch override
  --plan <path|id>    Explicit plan file or plan ID override
  --since <duration>  Override activity window (default 14d)
  --json              JSON output
  --verbose           Show all items uncapped
  --debug-score       Show commit attribute tags

Output template

=== resume: feature-name (cross-repo workstream) ===

Workstream: feature-name
  plan:    .cursor/plans/feature_name.plan.md
  branches:
    repo-a:     feature/name (8 commits ahead of main, 2d idle)
    repo-b:     feature/name (3 commits ahead of main, 5d idle)
  scope:
    repo-a:     api/handlers/, internal/auth/
    repo-b:     internal/commands/

Status (from plan checklist):
  [x] Step one                           (repo-a)
  [ ] Step two                           (repo-a)  ← next
  [~] Deferred item                      (repo-b)  [deferred]
  1/3 done, 1 open, 1 deferred

Tried and failed (heuristic):
  * [repo-a] Reverted approach: direct storage (commit f8a3b21 reverted)

Blockers (from transcript / plan):
  * Open question in plan: "should keys be workspace-scoped?"
    .cursor/plans/feature_name.plan.md > L142

Next steps (in plan order):
  1. Step two  (repo-a)  ← next

Sections are omitted entirely when empty — no "(none detected)" filler.

Workspace storage

Resume relies on workspace-scoped storage (see kleio config workspace show). The workspace resolver auto-detects your workspace from IDE state, workspace files, or multi-repo directory structure. See the CLI reference for details on kleio init --workspace and kleio migrate workspace.