status

name	status
description	Live system dashboard — 12-dimension ouroboros score, verdict v3, agents, build health, recommendations

For system state collection, follow .claude/skills/_shared/data_collection.md. For output formatting conventions, follow .claude/skills/_shared/output_format.md.

Mode Detection

Check $ARGUMENTS:

• "full" → MODE=FULL (detailed per-dimension breakdown + history + recommendations)
• "quick" or empty → MODE=QUICK (compact 20-line dashboard)
• "lang:ru" or "lang:en" → set language (read from .claude/skills/tri/lang.md, default: ru)

Data Collection

Run ALL of these in parallel (they are independent):

# 1. Ouroboros state
cat .trinity/ouroboros_state.json 2>/dev/null || echo '{"cycle":0}'

# 2. Verdict score (live — 12 dimensions)
zig-out/bin/tri verdict --explain 2>&1

# 3. Ouroboros status
zig-out/bin/tri ouroboros status 2>&1

# 4. Build health
zig build --summary none 2>&1; echo "EXIT:$?"

# 5. Git state
git status --porcelain | wc -l
git log --oneline -3

# 6. Agents alive
pgrep -la ralph-agent 2>/dev/null; pgrep -la tri-bot 2>/dev/null; pgrep -la trinity-mcp 2>/dev/null

# 7. Verdict history trend
cat .trinity/verdict_history.json 2>/dev/null | tail -c 500

# 8. Experience count
find .trinity/experience -name '*.json' 2>/dev/null | wc -l

# 9. Patent/IP status
cat .trinity/patent/status.json 2>/dev/null

QUICK Mode Output (default)

Compact dashboard, ~25 lines:

🐍 OUROBOROS v3 — HUNGRY SNAKE
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

  Score:      {score}/100 {emoji} {level}
  Cycle:      {cycle} | Strategy: {strategy}
  Stagnation: {stag}/3
  Build:      {pass|fail}
  Agents:     {list or "none"}
  Experience: {N} episodes

  TIER 1 — CRITICAL (50%)
    BUILD       {score}/100
    TEST_PASS   {score}/100
    TEST_COVER  {score}/100

  TIER 2 — CODE HEALTH (30%)
    TODO_DEBT   {score}/100
    GOD_FILES   {score}/100
    DEAD_CODE   {score}/100
    DUPLICATION {score}/100
    SPEC_GAP    {score}/100

  TIER 3 — EVOLUTION (20%)
    RESEARCH    {score}/100
    TOKEN_COST  {score}/100
    ENERGY      {score}/100

  Weakest:    {dimension} ({dim_score}/100) ← next target
  Rx:         {prescription one-liner}

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
  {timestamp} | tri ouroboros

FULL Mode Output

Everything from QUICK, plus:

Per-Dimension Breakdown

Table with all 12 dimensions grouped by tier, their scores, status (STRONG/OK/WEAK/CRITICAL), weight, and reason.

Score History

Last 5 entries from .trinity/verdict_history.json as mini sparkline or table:

  History: 99.4 → 27 → 35 → 42 → ...
           ↓72   ↑8   ↑7

Patent/IP Status

Show 4 discoveries from .trinity/patent/status.json:

  IP PROTECTION
    ternary-resonance-law    DOI ✅  Patent ❌
    square-attention         DOI ✅  Patent ❌
    0-dsp-fpga-inference     DOI ✅  Patent ❌
    self-evolving-ouroboros   DOI ✅  Patent ❌

Recommendations

Based on current state, generate 3 actionable recommendations:

1. If score < 30: "DISASTER. Run tri ouroboros --cycles 5 immediately"
2. If score 30-49: "GARBAGE. Fix BUILD first: tri ouroboros --cycles 3 --dimension BUILD"
3. If score 50-69: "MEDIOCRE. Target weakest: tri ouroboros --dimension {weakest}"
4. If score 70-89: "SOLID. Push to LEGENDARY: tri ouroboros --cycles 5"
5. If score >= 90 but weak dims exist: "Almost! {dim} still at {score}. tri ouroboros --dimension {dim}"
6. If all dims >= 95: "LEGENDARY. Snake is resting. File patents: tri patent status"
7. If build broken: "URGENT: Build broken. Fix compilation first"
8. If stagnation >= 2: "Strategy rotation imminent"

Git Activity

Last 3 commits with timestamps.

Next Actions

Bullet list of concrete tri commands to run based on the data above.

Language

For language detection and translations, follow .claude/skills/_shared/language.md. Default: ru. Technical terms stay in English.

Step: Telegram Broadcast

After rendering the dashboard, send a summary to Telegram.

Set TG_TEXT to the ouroboros summary (score, level, weakest dimension, 2-3 sentences in Russian, no mood signature). Set TG_MODE=dedup, TG_DEDUP_FILE=.trinity/tg_dedup_status.hash. Then execute the shared Telegram template from .claude/skills/_shared/telegram.md.

Step: Auto-Action (after dashboard + Telegram)

After rendering the dashboard and sending Telegram, execute the top-priority action automatically. Do NOT just list recommendations — ACT on the highest priority one.

Priority Cascade (first match wins):

1. Build broken → Run zig build 2>&1, diagnose, fix the error, verify
2. Dirty files > 50 → Group by directory, create conventional commits (skip .env, credentials, >1MB)
3. Score dropped ≥5 points → Run tri ouroboros --cycles 1 --dimension {weakest} to recover
4. GOD_FILES < 60 → Pick the largest god file, split it (max 1 file per cycle)
5. Weakest dimension < 70 → Run targeted fix for that dimension
6. Stagnation ≥ 2 → Run tri ouroboros --cycles 3 to break plateau
7. All healthy (score ≥ 90, no weak dims) → Print "✅ Система в φ-гармонии. Действий не требуется."

Rules:

• Only ONE action per /status cycle
• Report what was done in 1-2 sentences after the action
• If the action fails, report failure — do NOT retry in same cycle
• NEVER force-push, NEVER commit .env/credentials
• NEVER delete running services

Step: GitHub Issue Update (after action)

After auto-action completes, update the relevant GitHub issue with real-time status.

# Find the active status tracking issue (or use #357 as default farm tracker)
TRACKER=$(gh issue list -R gHashTag/trinity --label "status:in-progress" --json number --limit 1 -q '.[0].number' 2>/dev/null || echo "357")

# Post status comment with score, action taken, and current state
gh issue comment "$TRACKER" --body "## 🐍 Ouroboros Status — $(date +%Y-%m-%d\ %H:%M)

**Score**: {score}/100 ({level})
**Weakest**: {dimension} ({dim_score}/100)
**Action**: {what was done}
**Result**: {outcome}
**Dirty**: {count} files
**Farm**: {active} services, leader {name} PPL={ppl}" 2>/dev/null

Rules:

• ALWAYS post to GitHub after every /status action — this is the thought graph protocol
• If no in-progress issue exists, use #357 (farm tracker) as default
• Comment format must follow Trinity Protocol v2 (emoji + agent + timestamp)
• Status changes on issues must be visible in realtime — not batched

Session Memory

After each /status run, note key metrics for next session:

• Current score and level
• Which dimensions improved/regressed
• What recommendations were given
• What action was taken and its result
• Any anomalies (score drops, build failures, stagnation)

This builds context for the next conversation's /status call.