Every Story Should Be Resumable

2 commands · about 5 minutes · no migrationRun them in a repository. Choose your story from the guided picker after init. Setup is interactive.

Works with Jira, GitHub, Claude Code, and Codex. Your story context remains readable Markdown.

The code is there. The reason behind it isn't.

You return to a story—or hand it to a teammate—after several AI sessions. Jira describes the request and GitHub shows the changes, but neither tells you what was tried, what failed, what evidence changed the plan, what remains uncertain, or what should happen next. The next person or AI session starts by reconstructing the previous one.

See a resumable story

This public-safe excerpt and lifecycle recording both use Jesse's real CLI-only SCRUM-453 work record. They show what the story retained and what happened when a later session picked it up. This is founder dogfood evidence, not a customer testimonial.

Keep the configured archive directory outside Git managementSource: SCRUM-453 · CLI story record

Problem preserved

Git behavior around the configured archive could depend on .gitignore, so Imdone might still manage archived files that were unignored, explicitly allowed, or already tracked.

Scope corrected

The boundary applies to Imdone-managed Git operations. It must leave ordinary user-run Git, repository configuration, and existing tracking state unchanged.

Decision recorded

Source inspection mapped the shared Git path-selection boundary across status, diff, stash, stage, commit, removal, and reset. The plan centralizes the archive exclusion there and preserves provider-produced archive files through pull reconciliation.

Next action explicit

Begin with failing real-Git coverage for ignored, unignored, explicitly allowed, and already-tracked archive directories before changing production code.

A later AI session starts at the right next step

First session records Phase 1 complete at 35/35 focused tests. In a fresh run, the agent reads SCRUM-453’s story record, recovers the configuration-driven decision and Phase 2 test plan, starts that exact failing test, and finishes Phase 2 at 32/32 focused tests. One founder dogfood run; excerpts are edited only for time.

Full implementation continued later. This clip proves Phase 2 resumption, not completion of the entire story.

One story, resumed

Try it on one real story.

Use a real, active Jira or GitHub story that is likely to be interrupted. HDD Light keeps the problem, outcome, evidence, decisions, and next step with fewer artifacts.

Try it yourself on one real story

Two commands. About five minutes to start. No migration. Ask a teammate to test the handoff only after you have a record to share.

Copy the two-command test

Choose one active Jira or GitHub story from the guided picker after you run both commands in your repository.

Works on macOS, Linux, and Windows with Node.js and Git installed.

You will leave with a Markdown story record containing the current problem, decisions, evidence, uncertainty, and next action. imdone init performs the initial pull. Adding --resumability-test opens the guided story picker afterward. The guided path stays local; nothing is pushed until you run imdone push. Anonymous usage analytics do not include your story key or story contents.

npm install -g imdone-cli
imdone init --resumability-test

The --resumability-test flag offers the guided picker after init. Only the story you select is included in this resumability test. Its key stays in personal local state and is never sent with anonymous usage analytics.

Already initialized?

Start the same 5-minute test and choose a story from your searchable list:

See the complete eight-step HDD Light path

Choose one of your own stories
Pick real, active work rather than a sample generated by imdone.
Initialize and pull the story
imdone init --resumability-test performs the initial pull, then offers the guided story picker using the Jira or GitHub issues already in your repository.
Start HDD Light
Accept the five-minute offer and select a story. imdone enrolls it and starts the existing HDD Light path. If you skip the offer, continue later with imdone hdd --light --resumability-test; the command opens the same searchable picker.
Define the smallest outcome
Confirm the problem, what would be observably better, how feedback will be collected, and what evidence will show the result.
Approve a compact plan
Name the first boundary, important constraints, likely failure mode, and expected evidence before product-code changes begin.
Work and preserve progress
Keep meaningful decisions, blockers, corrections, completed work, evidence, and next steps with the story.
Deploy and close the loop
Record where the change is live, capture the result or feedback, and decide whether to close, iterate, or pivot.
Push the useful context and resume later
When you decide the record is ready to share, run imdone push STORY-123. Then test whether a later human or AI session can continue without reconstructing the work.

Working on something high-risk or ambiguous?

Use imdone hdd STORY-123 for high-risk, ambiguous, cross-team, provider-sync-sensitive, or implementation-heavy work. HDD Light is the recommended first path for a low-risk story.

Did the story become easier to resume?

In a later session, run imdone hdd, choose the enrolled story from the searchable picker, and confirm whether this is a real resumption. The CLI records that moment and starts the resumed session without an outcome questionnaire.

On the following launch, the CLI asks one question about the session that already happened. You do not need to return to this page or reuse the enrollment flag. Your answer is saved locally first and sent only when anonymous analytics are enabled.

Did imdone make this story materially easier to resume?

Give this to one teammate later

After your first run, test whether the record carries the work without replaying the prior sessions.

Resume STORY-123 from its story record without the previous AI transcript. Tell me what was clear, what was missing, and whether the next action was obvious.

When one story works, test the workflow with a team.

Choose how to continue after the one-story test.

Self-serve software

Start on your own

Use Free, $9 Personal, or $18-per-user Team without a guided pilot. Keep using the same story workflow at your own pace.

See pricing

Optional guided team pilot

Test continuity with real team work

Run a structured 30-day pilot using your existing Jira workflow and coding agents. It includes Team software access during the pilot and guided evaluation—not payment for CLI access alone.

3–6 developers
Team software access during the pilot
One active team workflow
A defined set of stories
Guided setup and team onboarding
Weekly evidence reviews
Findings and rollout recommendation

30 days · $1,500

Discuss a team pilot

Opens an email with four short qualification prompts. You choose what to send.

Every story should be resumable.

Keep the decisions, failed attempts, evidence, blockers, and next step behind a Jira or GitHub story so you, a teammate, or another AI agent can continue without reconstructing what happened.

The code is there. The reason behind it isn't.

See a resumable story

Problem preserved

Scope corrected

Decision recorded

Next action explicit

A later AI session starts at the right next step

Try it on one real story.

Try it yourself on one real story

Copy the two-command test

Already initialized?

Choose one of your own stories

Initialize and pull the story

Start HDD Light

Define the smallest outcome

Approve a compact plan

Work and preserve progress

Deploy and close the loop