RavBogard/dan:status

dan:status

Show the current project state: phase, plan, progress, blockers, recent decisions, and a suggested next action. This is the only skill Claude can auto-invoke (disable-model-invocation is false) to proactively check project status.

What It Does

1. Reads STATE.md for current position and progress
2. Reads ROADMAP.md for phase/plan overview
3. Reads recent SUMMARY.md files for context
4. Displays:
   • Current phase and plan
   • Progress bar and percentage
   • Active blockers or concerns
   • Recent decisions
   • Suggested next action (e.g., "Run dan:apply on 01-03-PLAN.md")

Execution Mode

Runs in-session (no agent spawning). Read-only operation.

CLI Tools

State operations use the DAN CLI:

node "$HOME/.claude/dan/bin/dan-tools.cjs" --cwd $PROJECT_DIR state read
node "$HOME/.claude/dan/bin/dan-tools.cjs" --cwd $PROJECT_DIR milestone progress
node "$HOME/.claude/dan/bin/dan-tools.cjs" --cwd $PROJECT_DIR milestone status
node "$HOME/.claude/dan/bin/dan-tools.cjs" --cwd $PROJECT_DIR session next-action

<execution_flow>

<step name="gather_state" priority="first"> ## Step 1: Gather State

Read all state sources needed for the status display.

1a. Read STATE.md

STATE_JSON=$(node "$HOME/.claude/dan/bin/dan-tools.cjs" --cwd "$PROJECT_DIR" state read)

Extract from STATE_JSON:

• phase: current phase number (e.g., "3 of 5")
• plan: current plan position (e.g., "2 of 4 in current phase")
• status: current status (e.g., "in_progress", "verifying", "blocked")
• last_activity: most recent activity description
• blockers: list of active blockers/concerns
• decisions: list of accumulated decisions

1b. Get milestone progress

PROGRESS_JSON=$(node "$HOME/.claude/dan/bin/dan-tools.cjs" --cwd "$PROJECT_DIR" milestone progress)

Extract from PROGRESS_JSON:

• total_phases: total number of phases
• completed_phases: number of completed phases
• total_plans: total plans across all phases
• completed_plans: number of completed plans
• percent: overall progress percentage

1c. Get milestone status

STATUS_JSON=$(node "$HOME/.claude/dan/bin/dan-tools.cjs" --cwd "$PROJECT_DIR" milestone status)

Extract from STATUS_JSON:

• milestone: milestone name
• current_phase: name of current phase
• pipeline_stage: current pipeline stage if milestone is running (e.g., "apply", "verify")
• phases_remaining: list of remaining phase numbers </step>

<step name="determine_next_action"> ## Step 2: Determine Next Action

Call the session module to compute the recommended next action based on the 8-priority state machine:

NEXT_JSON=$(node "$HOME/.claude/dan/bin/dan-tools.cjs" --cwd "$PROJECT_DIR" session next-action)

Extract from NEXT_JSON:

• action: human-readable description (e.g., "Apply next plan", "Verify phase", "Resume paused task")
• skill: the DAN skill to invoke (e.g., "dan:apply", "dan:verify", "dan:milestone")
• args: arguments to pass to the skill (e.g., plan path, phase number)

The priority order is:

1. Paused session exists -> resume
2. Plan IN_PROGRESS -> continue apply
3. All plans done, no verification -> verify
4. Verification has gaps -> bugsweep
5. Phase done, more phases remain -> next phase
6. All phases done -> complete
7. No plans exist -> plan
8. No research exists -> research </step>

<step name="display_status"> ## Step 3: Display Status

Format and display the status report using the gathered data.

Phase position

Show the current phase with its name:

## Project Status

**Phase {completed_phases + 1} of {total_phases}:** {current_phase name}

Progress bar

Render the progress bar using the same format as the CLI (10-character width):

Progress: [{filled}{empty}] {percent}%
          [{completed_plans}/{total_plans} plans complete]

Where filled = number of filled blocks proportional to percent, empty = remaining blocks.

Example: Progress: [######....] 60% [9/15 plans complete]

Current activity

<if condition="status indicates a plan is IN_PROGRESS"> ``` **Current plan:** {plan identifier} **Status:** {status description} **Last activity:** {last_activity text} ``` </if> <if condition="status indicates milestone pipeline is running"> ``` **Pipeline stage:** {pipeline_stage} (phase {current phase}) **Last activity:** {last_activity text} ``` </if> <if condition="status is idle or complete"> ``` **Status:** {status} **Last activity:** {last_activity text} ``` </if>

Blockers and concerns

<if condition="blockers list is non-empty"> ``` ### Blockers

{numbered list of each blocker/concern from STATE.md}

</if>

<if condition="blockers list is empty">

Blockers

None.

</if>

### Recent decisions

Show the last 3 decisions from the STATE.md decisions list:

Recent Decisions

{last 3 decisions, most recent first}

If no decisions exist, show "No decisions recorded yet."

### Suggested next action

Highlight the recommended next action from Step 2:

Next Action

{action description} Run: /dan:{skill} {args}

If the project is complete (no next action), show:

Next Action

Project milestone is complete. No further actions needed.

</step>

<step name="format_output">
## Step 4: Format Output

Assemble the complete status display as a single markdown block. Use clean formatting with headers and consistent spacing.

Full output template:

```markdown
## Project Status

**Phase {N} of {total}:** {phase name}

Progress: [{bar}] {percent}%  [{completed_plans}/{total_plans} plans]

**Status:** {status}
**Last activity:** {last_activity}

### Blockers

{blockers or "None."}

### Recent Decisions

{last 3 decisions or "No decisions recorded yet."}

### Next Action

> **{action}**
> Run: `/dan:{skill} {args}`

This skill is read-only. It does NOT modify STATE.md, ROADMAP.md, or any other file. It only reads and displays. </step>

Constraints

• Read-only: This skill never modifies any project state. It only reads and displays.
• Auto-invocable: disable-model-invocation: false allows Claude to proactively call this skill to check status when context suggests it would be helpful.
• No agent spawning: Runs entirely in-session. No subagents needed for read-only display.
• Concise output: The status display should fit in a single screen (roughly 20-30 lines). Do not dump raw JSON or full STATE.md contents.
• Graceful degradation: If any CLI call fails (e.g., no STATE.md exists yet), show what is available and note what is missing rather than failing entirely.

</execution_flow>