roadmap（roadmap 命令行）

/roadmap - Active Roadmap View

Usage

/roadmap                         # All active plans sorted by priority
/roadmap --type=workflow         # Filter by domain prefix (matches workflow:*)
/roadmap --type=workflow:migration  # Filter by exact compound type
/roadmap --type=lego             # Filter by domain prefix (matches lego:*)
/roadmap --priority=P1           # Show only P1 items
/roadmap --prefix=APIP           # Filter by story prefix

Multiple filters can be combined: /roadmap --priority=P1 --type=workflow

Plan Type Schema

Plan types use a compound domain:subtype key. Known values:

When filtering with --type=X:

• If X contains : → exact match on full compound type (e.g. --type=workflow:migration)
• If X has no : → prefix match on domain (e.g. --type=workflow matches workflow:migration, workflow:feature, etc.)

Execution

Step 1 — Parse Arguments

Parse the user's input for optional filters:

Step 2 — Query KB

Preferred: Call kb_list_plans for each active status: in-progress, stories-created, draft, active, accepted. Make the five calls in parallel.

• Pass priority and story_prefix filters directly to each call if provided.
• Pass plan_type only if --type=X contains : (exact match). If X has no :, fetch unfiltered and apply prefix match client-side after merging.
• Use limit: 50 for each call.

Fallback (if MCP tools unavailable): Query the database directly:

const { Client } = require('./apps/api/knowledge-base/node_modules/pg')
const c = new Client({
  connectionString: 'postgresql://kbuser:TestPassword123!@localhost:5433/knowledgebase',
})

SQL to fetch plans:

SELECT plan_slug, title, LEFT(COALESCE(summary,''), 100) AS summary,
       plan_type, status, feature_dir, story_prefix, estimated_stories, priority, updated_at
FROM workflow.plans
WHERE status IN ('in-progress','stories-created','draft','active','accepted')
  AND deleted_at IS NULL
  [AND priority = $1]              -- if --priority filter
  [AND (plan_type = $2             -- if --type=X with ':'
        OR plan_type LIKE $2||':%')]  -- if --type=X without ':'
  [AND story_prefix = $3]          -- if --prefix filter
ORDER BY
  CASE status WHEN 'in-progress' THEN 1 WHEN 'stories-created' THEN 2 WHEN 'draft' THEN 3 END,
  priority,
  plan_slug;

Merge all results into a single list and sort by:

1. Priority (P1 first)
2. Status order: in-progress > stories-created > draft
3. Plan slug alphabetically

Step 3 — Format Output

Display results as a markdown table:

| Priority | Type | Status | Plan Slug | Description | Prefix | Stories | Updated |

Column formatting:

• Priority: P1-P5
• Type: full compound type value (e.g. workflow:migration, lego:feature)
• Status: as-is
• Plan Slug: backtick-wrapped slug
• Description: use summary field, truncate to 80 chars if needed (append "...")
• Prefix: story_prefix or --
• Stories: For in-progress and stories-created plans that have a featureDir:

- Use basename(featureDir) as the feature filter for kb_list_stories (e.g. "plans/future/platform/wint" → "wint") - Or query the DB: SELECT COUNT(*), COUNT(*) FILTER (WHERE state='completed') FROM public.stories WHERE feature=$1 AND deleted_at IS NULL - Display as completed/total (e.g. 3/20) - For draft plans or plans with no stories in DB, show estimated_stories or --

• Updated: relative date (e.g. "2h ago", "3d ago", "today")

Step 4 — Summary Line

After the table, output a one-line summary:

Roadmap: N active plans — X draft, Y accepted, Z stories-created, W in-progress

Count each status from the returned results.