bg-szy/TOP-SKILLS

BV - Beads Viewer

A graph-aware triage engine for Beads projects (.beads/beads.jsonl). Computes 9 graph metrics, generates execution plans, and provides deterministic recommendations. Human TUI for browsing; robot flags for AI agents.

Why BV vs Raw Beads

Capability	Raw beads.jsonl	BV Robot Mode
Query	"List all issues"	"List the top 5 bottlenecks blocking the release"
Context Cost	High (linear with issue count)	Low (fixed summary struct)
Graph Logic	Agent must compute	Pre-computed (PageRank, betweenness, cycles)
Safety	Agent might miss cycles	Cycles explicitly flagged

Use BV instead of parsing beads.jsonl directly. It computes graph metrics deterministically.

CRITICAL: Robot Mode for Agents

Never run bare bv. It launches an interactive TUI that blocks your session.

Always use --robot-* flags:

bv --robot-triage        # THE MEGA-COMMAND: start here
bv --robot-next          # Minimal: just the single top pick
bv --robot-plan          # Parallel execution tracks
bv --robot-insights      # Full graph metrics

The 9 Graph Metrics

BV computes these metrics to surface hidden project dynamics:

Metric	What It Measures	Key Insight
PageRank	Recursive dependency importance	Foundational blockers
Betweenness	Shortest-path traffic	Bottlenecks and bridges
HITS	Hub/Authority duality	Epics vs utilities
Critical Path	Longest dependency chain	Keystones with zero slack
Eigenvector	Influence via neighbors	Strategic dependencies
Degree	Direct connection counts	Immediate blockers/blocked
Density	Edge-to-node ratio	Project coupling health
Cycles	Circular dependencies	Structural errors (must fix!)
Topo Sort	Valid execution order	Work queue foundation

Two-Phase Analysis

BV uses async computation with timeouts:

• Phase 1 (instant): degree, topo sort, density
• Phase 2 (500ms timeout): PageRank, betweenness, HITS, eigenvector, cycles

Always check status field in output. For large graphs (>500 nodes), some metrics may be approx or skipped.

Robot Commands Reference

Triage & Planning

bv --robot-triage              # Full triage: recommendations, quick_wins, blockers_to_clear
bv --robot-next                # Single top pick with claim command
bv --robot-plan                # Parallel execution tracks with unblocks lists
bv --robot-priority            # Priority misalignment detection

Graph Analysis

bv --robot-insights            # Full metrics: PageRank, betweenness, HITS, cycles, etc.
bv --robot-label-health        # Per-label health: healthy|warning|critical
bv --robot-label-flow          # Cross-label dependency flow matrix
bv --robot-label-attention     # Attention-ranked labels

History & Changes

bv --robot-history             # Bead-to-commit correlations
bv --robot-diff --diff-since <ref>  # Changes since ref

Other Commands

bv --robot-burndown <sprint>   # Sprint burndown, scope changes
bv --robot-forecast <id|all>   # ETA predictions
bv --robot-alerts              # Stale issues, blocking cascades
bv --robot-suggest             # Hygiene: duplicates, missing deps, cycle breaks
bv --robot-graph               # Dependency graph export (JSON, DOT, Mermaid)
bv --export-graph <file.html>  # Self-contained interactive HTML visualization

Scoping & Filtering

bv --robot-plan --label backend              # Scope to label's subgraph
bv --robot-insights --as-of HEAD~30          # Historical point-in-time
bv --recipe actionable --robot-plan          # Pre-filter: ready to work
bv --recipe high-impact --robot-triage       # Pre-filter: top PageRank
bv --robot-triage --robot-triage-by-track    # Group by parallel work streams
bv --robot-triage --robot-triage-by-label    # Group by domain

Built-in Recipes

Recipe	Purpose
default	All open issues sorted by priority
actionable	Ready to work (no blockers)
high-impact	Top PageRank scores
blocked	Waiting on dependencies
stale	Open but untouched for 30+ days
triage	Sorted by computed triage score
quick-wins	Easy P2/P3 items with no blockers
bottlenecks	High betweenness nodes

Robot Output Structure

All robot JSON includes:

• data_hash - Fingerprint of beads.jsonl (verify consistency)
• status - Per-metric state: computed|approx|timeout|skipped
• as_of / as_of_commit - Present when using --as-of

--robot-triage Output

{
  "quick_ref": { "open": 45, "blocked": 12, "top_picks": [...] },
  "recommendations": [
    { "id": "bd-123", "score": 0.85, "reason": "Unblocks 5 tasks", "unblock_info": {...} }
  ],
  "quick_wins": [...],
  "blockers_to_clear": [...],
  "project_health": { "distributions": {...}, "graph_metrics": {...} },
  "commands": { "claim": "bd claim bd-123", "view": "bv --bead bd-123" }
}

--robot-insights Output

{
  "bottlenecks": [{ "id": "bd-123", "value": 0.45 }],
  "keystones": [{ "id": "bd-456", "value": 12.0 }],
  "influencers": [...],
  "hubs": [...],
  "authorities": [...],
  "cycles": [["bd-A", "bd-B", "bd-A"]],
  "clusterDensity": 0.045,
  "status": { "pagerank": "computed", "betweenness": "computed", ... }
}

jq Quick Reference

bv --robot-triage | jq '.quick_ref'                        # At-a-glance summary
bv --robot-triage | jq '.recommendations[0]'               # Top recommendation
bv --robot-plan | jq '.plan.summary.highest_impact'        # Best unblock target
bv --robot-insights | jq '.status'                         # Check metric readiness
bv --robot-insights | jq '.cycles'                         # Circular deps (must fix!)
bv --robot-label-health | jq '.results.labels[] | select(.health_level == "critical")'

Agent Workflow Pattern

# 1. Start with triage
TRIAGE=$(bv --robot-triage)
NEXT_TASK=$(echo "$TRIAGE" | jq -r '.recommendations[0].id')

# 2. Check for cycles first (structural errors)
CYCLES=$(bv --robot-insights | jq '.cycles')
if [ "$CYCLES" != "[]" ]; then
  echo "Fix cycles first: $CYCLES"
fi

# 3. Claim the task
bd claim "$NEXT_TASK"

# 4. Work on it...

# 5. Close when done
bd close "$NEXT_TASK"

TUI Views (for Humans)

When running bv interactively (not for agents):

Key	View
l	List view (default)
b	Kanban board
g	Graph view (dependency DAG)
E	Tree view (parent-child hierarchy)
i	Insights dashboard (6-panel metrics)
h	History view (bead-to-commit correlation)
a	Actionable plan (parallel tracks)
f	Flow matrix (cross-label dependencies)
]	Attention view (label priority ranking)

Integration with bd CLI

BV reads from .beads/beads.jsonl created by the bd CLI:

bd init                    # Initialize beads in project
bd create "Task title"     # Create a bead
bd list                    # List beads
bd ready                   # Show actionable beads
bd claim bd-123            # Claim a bead
bd close bd-123            # Close a bead

Integration with Agent Mail

Use bead IDs as thread IDs for coordination:

file_reservation_paths(..., reason="bd-123")
send_message(..., thread_id="bd-123", subject="[bd-123] Starting...")

Graph Export Formats

bv --robot-graph                              # JSON (default)
bv --robot-graph --graph-format=dot           # Graphviz DOT
bv --robot-graph --graph-format=mermaid       # Mermaid diagram
bv --robot-graph --graph-root=bd-123 --graph-depth=3  # Subgraph
bv --export-graph report.html                 # Interactive HTML

Time Travel

Compare against historical states:

bv --as-of HEAD~10                    # 10 commits ago
bv --as-of v1.0.0                     # At tag
bv --as-of "2024-01-15"               # At date
bv --robot-diff --diff-since HEAD~30  # Changes in last 30 commits

Common Pitfalls

Issue	Fix
TUI blocks agent	Use --robot-* flags only
Stale metrics	Check status field, results cached by data_hash
Missing cycles	Run --robot-insights, check .cycles
Wrong recommendations	Use --recipe actionable to filter to ready work

Performance Notes

• Phase 1 metrics (degree, topo, density): instant
• Phase 2 metrics (PageRank, betweenness, etc.): 500ms timeout
• Results cached by data_hash
• Prefer --robot-plan over --robot-insights when speed matters