Lockstep Core v0.2.1 — Patch Release

Chain-based project tracking for Human+AI collaboration. An MCP server that gives your AI assistant persistent memory across sessions.

What's in this patch

Startup diagnostics. When we tested v0.2.0 from a fresh user perspective (not the creator's), the server started with an empty dashboard and no errors. Data wasn't lost — the server had silently resolved to a default empty directory instead of the configured one. That kind of silent failure is unacceptable.

Now every server startup logs its data directory and how it was resolved (environment variable, config file, or default fallback). The get_dashboard response includes a diagnostics block so you can verify your configuration is working. If the server falls back to a default empty directory, it warns you.

UPGRADE.md. Step-by-step migration guide for v0.1.0 → v0.2.0 upgrades covering three configuration methods, the known MCPB interpolation bug and its workaround, and troubleshooting for the "my data appears lost" scenario.

Known Issue: MCPB user_config interpolation

There is a confirmed bug in MCPB where ${user_config.*} values in the manifest env block are not interpolated into the spawned process environment. This means the LOCKSTEP_DATA_DIR env var set through Claude Desktop's settings UI may not reach the server. This has been reported upstream.

Workaround: Create a config.json file in your data directory:

// Create src/config.json adjacent to server.py
{
  "data_dir": "/path/to/your/lockstep/data"
}

This bypasses MCPB interpolation entirely. The server checks config.json before falling back to the default directory. See UPGRADE.md for full details.

Upgrading

If you're already on v0.2.0, this is a drop-in replacement — no data migration needed.

If upgrading from v0.1.0, see UPGRADE.md for the complete migration guide.

Install

MCPB Bundle (Recommended)

1. Download lockstep-core.mcpb from the assets below
2. Open it with Claude Desktop (double-click or drag in)
3. When prompted, set your data directory
4. Call get_dashboard — check the diagnostics block to confirm your path

Manual Setup

See the README for platform-specific instructions.

Full changelog: CHANGELOG.md

License

GPL v3 — free to use, modify, and distribute. Modifications must stay open source.

Built by Dandelion Rose Group — proving that neurodivergent minds are uniquely wired for AI partnership.

Lockstep Core v0.2.0 — "Tickets Grow Up"

Chain-based project tracking for Human+AI collaboration. An MCP server that gives your AI assistant persistent memory across sessions.

What's new in v0.2.0

Tickets become chains. The biggest gap in v0.1.0 was that tickets couldn't grow. Now they can — promote_ticket turns any standalone ticket into a full chain when the work outgrows its original scope, with automatic scanning for related tickets to nest.

Chain types are YAML, not code. Four built-in templates (full-funnel, enhancement, refactor, bug-fix) ship out of the box. Need a custom workflow? Drop a .yaml file in templates/ — no code changes, no version bump.

Progressive disclosure. Early-phase work shows only what matters. Discovery sessions surface completion vision and tags; build phases expand to show everything. Less noise when you're exploring, full detail when you're executing.

Cross-platform support. Tested on macOS (ARM), Windows 11 Pro (x64), and Linux. 51 tests, zero platform-specific failures. The manifest now declares explicit compatibility and uses the uv server type for host-managed Python across all platforms.

Highlights

• Ticket promotion — tickets grow into chains with candidate scanning and nesting
• YAML-defined chain types — add new workflow types without touching code
• Progressive disclosure — per-phase field visibility reduces cognitive load
• Child chain spawning — cross-type forks that preserve lineage
• Advisory nudges — soft suggestions when tickets are ready for promotion
• Schema migration — v1 → v2 auto-migration with backup script
• Cross-platform — verified on macOS, Windows 11, and Linux
• 42 tools + 5 commands for full project lifecycle management

Install

MCPB Bundle (Recommended)

1. Download lockstep-core.mcpb from the assets below
2. Open it with Claude Desktop (double-click or drag in)
3. When prompted, choose a data directory (default: ~/.lockstep/data)

Manual Setup

See the README for platform-specific instructions (macOS, Windows, Linux).

Upgrading from v0.1.0

Your existing data works as-is — the server auto-migrates v1 files on read. For a clean migration:

python scripts/migrate_v1_to_v2.py ~/.lockstep/data

Full changelog: CHANGELOG.md

License

GPL v3 — free to use, modify, and distribute. Modifications must stay open source.

Built by Dandelion Rose Group — proving that neurodivergent minds are uniquely wired for AI partnership.

Lockstep Core v0.1.0 — First Public Release

Chain-based project tracking for Human+AI collaboration. An MCP server that gives your AI assistant persistent memory across sessions.

What's in this release

• 40 tools + 5 commands for full project lifecycle management
• Chain-based tracking — sessions link together so nothing gets lost between conversations
• Structured handoffs — decisions, context, and next steps transfer seamlessly
• Capacity tracking — growth stages that make invisible progress visible
• Advisory, not enforcing — the protocol flags and explains, never blocks
• Fully local — all data stored as YAML on your machine. No network. No telemetry. No data collection.

Install

1. Download lockstep-chain-protocol.mcpb from the assets below
2. Open Claude Desktop → Settings → Extensions
3. Click Install Extension and select the downloaded .mcpb file
4. When prompted, choose a data directory (default: ~/.lockstep/data)
5. Start a conversation and try: "Create a new initiative called 'My First Project' with the vision 'Learn how Lockstep works'"

Who is this for?

Anyone using AI coding assistants who wants structured session continuity. Especially useful for neurodivergent individuals who benefit from external scaffolding — Lockstep makes the invisible work of executive function visible and trackable.

License

GPL v3 — free to use, modify, and distribute. Modifications must stay open source.

Built by Dandelion Rose Group — proving that neurodivergent minds are uniquely wired for AI partnership.