Ticketing Insights

Ticketing Insights

Mode Declaration

This skill operates in REPORTING mode only.

Ticketing-insights is not a build skill and not a close-out skill. It reads data — from Linear, GitHub, and git history — and produces reports. It does not implement features, merge PRs, or modify worktrees.

Valid --mode values:

• deep-dive (default) — comprehensive daily work analysis
• close-day — end-of-day reconciliation (ModelDayClose YAML)
• project — project health dashboard
• velocity — milestone ETA calculations
• suggest — priority backlog recommendations
• pipeline — rework ratio, cycle time, CI stability
• github — PR counts, commit velocity, LOC metrics
• all — run every mode, produce combined report

First output line must always be:

[ticketing-insights] MODE: reporting | submode: <mode-value>

No tool calls, file reads, or bash commands may precede this output.

If an unrecognized --mode value is passed, emit:

ERROR: Unknown --mode value: <value>
Valid modes: deep-dive | close-day | project | velocity | suggest | pipeline | github | all

Then stop.

Unified analytics and reporting for the OmniNode platform. All reporting modes are subcommands of this single skill.

Quick Start

/ticketing-insights                           # Default: deep-dive mode
/ticketing-insights --mode close-day          # End-of-day reconciliation (ModelDayClose YAML)
/ticketing-insights --mode project MVP        # Quick health dashboard
/ticketing-insights --mode velocity MVP       # Milestone ETA calculations
/ticketing-insights --mode suggest            # Priority backlog recommendations
/ticketing-insights --mode pipeline           # Rework ratio, cycle time, CI stability
/ticketing-insights --mode github             # PR counts, commit velocity, LOC metrics
/ticketing-insights --mode all                # Run every mode, produce combined report

Modes

--mode deep-dive (default)

Generate a comprehensive daily work analysis. See Deep Dive Report section below.

--mode close-day

Auto-generates a ModelDayClose YAML from today's GitHub PRs, git activity, and invariant probes. Absorbed from the former close-day skill.

Invocation: /linear-insights --mode close-day [--date YYYY-MM-DD] [--dry-run]

Execution steps:

1. Pull merged PRs across all OmniNode-ai repos for the target date
2. Fetch active-sprint Linear plan
3. Build actual_by_repo grouped by repo with ticket references
4. Detect scope drift: PRs with no ticket ref → drift_detected entry
5. Run scripts/check_arch_invariants.py (shared with CDQA-07) to probe reducers/orchestrators
6. Detect golden-path progress by reading emitted_at from $ONEX_STATE_DIR/golden-path/TODAY/ artifact JSON files
7. Set unknown statuses to "unknown" and add entries to corrections_for_tomorrow
8. Validate against ModelDayClose.model_validate() (fails loudly on schema mismatch)
9. Write to $ONEX_CC_REPO_PATH/drift/day_close/YYYY-MM-DD.yaml or print with warning banner

ONEX_CC_REPO_PATH behavior: If not set, YAML is printed with a warning banner (file NOT written). If set and path exists, writes to the drift directory.

--mode project

Quick health dashboard for Linear projects — progress, velocity, blockers, ETA, and confidence. Absorbed from the former project-status skill.

Invocation: /linear-insights --mode project [PROJECT] [--all] [--blockers] [--risks] [--confidence] [--json] [--emit]

Supports --emit to relay workstream snapshots to Kafka via onex-linear-relay.

--mode velocity

Calculate project velocity from historical data and estimate milestone completion dates. Absorbed from the former velocity-estimate skill.

Invocation: /linear-insights --mode velocity [PROJECT] [--all] [--confidence] [--history] [--weighted] [--method METHOD] [--json]

Methods: simple (signal-grade), priority (signal), points (ETA-grade), labels (signal), cycle_time (ETA-grade).

--mode suggest

Priority backlog recommendations — highest priority unblocked issues with repo-based prioritization. Absorbed from the former suggest-work skill.

Invocation: /linear-insights --mode suggest [--count N] [--project PROJECT] [--repo REPO] [--label LABEL] [--json]

--mode pipeline

Pipeline health metrics — rework ratio, cycle time, CI stability, and feature velocity. Absorbed from the former pipeline-metrics skill.

Invocation: /linear-insights --mode pipeline [--since YYYY-MM-DD] [--ticket TICKET-ID] [--repo REPO] [--format table|json]

Metrics: Cycle Time (P50/P90), CI Clean Rate, Rework Cycles/Ticket, Feature Velocity, Skill Duration P50.

--mode github

GitHub repository statistics — PR counts, commit velocity, contributor activity, LOC metrics. Absorbed from the former gather-github-stats skill.

Invocation: /linear-insights --mode github [--github-only] [--local-only] [--include-local-loc] [--include-private] [--output PATH]

--mode all

Runs every mode sequentially and produces a combined report. Useful for comprehensive end-of-sprint summaries.

When to Use

• End of day wrap-up: --mode deep-dive or --mode close-day
• Sprint planning: --mode velocity for capacity planning
• Milestone tracking: --mode project for data-driven ETAs
• What to work on next: --mode suggest for prioritized backlog
• Pipeline health check: --mode pipeline for rework and CI metrics
• Repo statistics: --mode github for PR throughput and commit velocity
• Retrospectives: --mode all for comprehensive analysis

Deep Dive Report

Generates a comprehensive analysis of work completed in a specified time period. Format matches the established deep dive pattern (see $ONEX_STATE_DIR/deep-dives/ for archive).

Usage

# Today's deep dive (display only) — auto-discovers all active repos
${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/deep-dive

# Specific date (auto-discovers repos with activity on that date)
${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/deep-dive --date 2025-12-09

# Last N days (for weekly summary)
${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/deep-dive --days 7

# Save to default directory (omni_save)
${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/deep-dive --save

# Save to custom directory
${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/deep-dive --save --output-dir ~/reports

# JSON output for processing
${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/deep-dive --json

# Analyze specific repos only (overrides auto-discovery)
${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/deep-dive --repos omnibase_core,omniclaude

Configuration

Output Directory (where reports are saved with --save):

Filename Pattern: {MONTH}_{DAY}_{YEAR}_DEEP_DIVE.md

• Example: DECEMBER_13_2025_DEEP_DIVE.md

Deep Dive Format

The deep dive follows a structured format with these sections:

1. Executive Summary

• Velocity Score: 0-100 based on commit volume, PRs merged, issues completed
• Effectiveness Score: 0-100 based on strategic value of work completed
• Overall Assessment: 2-3 sentence summary of the day

2. Repository Activity Overview

• Commits per repository
• PRs merged per repository
• Files changed and lines added/deleted
• Focus areas

3. Major Components & Work Completed

For each PR merged:

• Status, Impact level (Critical/High/Medium/Low)
• Files changed, lines added/deleted
• Description of work
• Key components/features
• Linear tickets addressed
• Significance statement

4. Detailed Commit Analysis

• Commits grouped by category (Contracts, Runtime, CI, etc.)
• Key individual commits with file counts

5. Metrics & Statistics

• Total commits, PRs, files changed
• PR statistics table
• Linear ticket progress (closed/in-progress)
• Code quality metrics

6. Work Breakdown by Category

• Percentage breakdown of work types
• Time/effort allocation

7. Key Achievements

• Bullet points of major accomplishments
• Milestone progress

8. Challenges & Issues

• Technical challenges encountered
• Process observations

9. Velocity Analysis

• Positive/negative velocity factors
• Velocity score justification

10. Effectiveness Analysis

• High-value work identified
• Strategic impact assessment
• Effectiveness score justification

11. Lessons Learned

• Key takeaways from the day
• Insights for future work

12. Next Day Preview

• Expected focus areas for tomorrow
• Upcoming priorities

13. Appendix

• Complete commit log with timestamps
• PR details

Example Output

# December 13, 2025 - Deep Dive Analysis

**Date**: Friday, December 13, 2025
**Week**: Week of December 9-13, 2025
**Day of Week**: Friday

---

## Executive Summary

**Velocity Score**: 85/100
**Effectiveness Score**: 90/100

**Overall Assessment**: Strong day with 12 issues completed across MVP and Beta milestones.
Focus on code quality improvements and deprecation fixes. 6 PRs merged with comprehensive
test coverage improvements.

---

## Repository Activity Overview

### omnibase_core
**Total Commits**: 24
**PRs Merged**: 4 (PRs #184-188)
**Files Changed**: 156
**Lines Changed**: +8,234 / -2,891
**Focus Areas**: Deprecation fixes, purity violations, structured logging

### omnibase_spi
**Total Commits**: 8
**PRs Merged**: 2 (PRs #39-40)
**Files Changed**: 23
**Lines Changed**: +1,456 / -892
**Focus Areas**: EventBus protocol cleanup, test coverage

...

Velocity Estimate

Calculates project velocity and estimates time to completion for milestones.

Usage

# Estimate for MVP project
${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/velocity-estimate --project "MVP"

# All milestones overview
${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/velocity-estimate --all

# Include confidence intervals
${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/velocity-estimate --project "MVP" --confidence

# JSON output
${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/velocity-estimate --project "Beta" --json

Metrics Calculated

Velocity Calculation

Velocity is calculated using a weighted rolling average:

• Last 7 days: 50% weight (recent performance)
• Last 14 days: 30% weight (short-term trend)
• Last 30 days: 20% weight (baseline)

This balances recent momentum with historical patterns.

Example Output

# Velocity Report: MVP - OmniNode Platform Foundation

## Current Status
- **Total Issues**: 71
- **Completed**: 28 (39%)
- **In Progress**: 5
- **Backlog**: 38

## Velocity Metrics
- **7-day velocity**: 2.3 issues/day
- **14-day velocity**: 1.8 issues/day
- **30-day velocity**: 1.5 issues/day
- **Weighted velocity**: 2.0 issues/day

## Completion Estimate
- **Remaining Issues**: 43
- **Estimated Days**: 21.5 days
- **Target Date**: 2026-01-03
- **Confidence**: Medium (variance: 0.4)

## Velocity Trend
[Chart showing 30-day velocity trend]

Week of 12/02: ████████░░ 1.6/day
Week of 12/09: ██████████ 2.3/day (current)

## Risk Factors
- 5 issues blocked/waiting
- 2 urgent issues in backlog
- Holiday period may reduce velocity

Estimation Accuracy (Factory Telemetry)

Three-layer factory telemetry that parses the deep dive archive for historical velocity, effectiveness, PR throughput, and fix-vs-feature trends. Requires python3.

Three Data Layers

Layer 1 (Deep Dive Archive): Parsed at script time from $ONEX_STATE_DIR/deep-dives/. Handles three format eras (Dec 2025, Feb 2026, Mar 2026) with null-not-zero for missing data. This is the primary data source.

Layer 2 (GitHub PRs): The script outputs gh pr list commands with date-window filtering. The agent executes these to collect merged PR data for reconciliation.

Layer 3 (Linear Done): The script outputs Linear MCP call instructions (state="Done"). The agent executes these to compare archive evidence against Linear's Done state.

Usage

# All data + agent instructions
${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/estimation-accuracy

# Last 7 days
${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/estimation-accuracy --week

# Date range
${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/estimation-accuracy --from 2026-03-01 --to 2026-03-05

# Graphable JSON output
${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/estimation-accuracy --json

# Step-by-step agent instructions
${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/estimation-accuracy --generate

CLI Flags

Fix-vs-Feature Tracking

PRs are classified by deep dive category: capability/governance/observability count as feature, correctness/churn count as fix, docs are excluded from both. Weekly aggregation computes fix ratio per ISO week. Trend analysis requires >= 4 weeks of data; labels are declining, stable, increasing, or insufficient_data.

JSON Output Keys

• meta -- period, deep dive count, generation timestamp
• time_series -- per-day entries with velocity, effectiveness, prs_merged, categories, parse_quality
• fix_vs_feature -- weekly aggregation with trend_slope and trend_label
• throughput_by_repo -- per-repo PR totals and daily averages (Era B/C only)
• reconciliation -- extracted ticket IDs and instructions for agent to fill GitHub/Linear data

See script header comments for full schema.

Reconciliation

The reconciliation view computes a three-way set comparison using ticket identifiers: archive IDs (A), Linear Done IDs (B), and GitHub PR IDs (C). The script computes set A; the agent fills B and C after executing the Layer 2/3 instructions, then computes overlap, gap ratio, and tickets shipped but not closed in Linear.

Data Sources

• Deep dive archive: $ONEX_STATE_DIR/deep-dives/*_DEEP_DIVE.md (Layer 1, parsed by script)
• GitHub CLI: gh pr list commands (Layer 2, agent-executed)
• Linear MCP: tracker.list_issues (Layer 3, agent-executed)
• tracker.list_projects - Project metadata
• tracker.get_project - Project details

No external databases or caches are used - all calculations are done on fresh Linear data.

Implementation Notes

For General-Purpose Agent Dispatch

These skills are designed to be invoked by general-purpose agents:

Task(
  subagent_type="general-purpose",
  description="Generate daily work report",
  prompt="Generate a daily work report for the last 24 hours.

    Use Linear MCP tools:
    1. tracker.list_issues with assignee='me' and updatedAt='-P1D'
    2. Categorize by status (Done, In Progress, Backlog)
    3. Extract PR links from attachments
    4. Group by repository from labels

    Format as markdown with:
    - Summary stats
    - Completed work section with PR links
    - In Progress section
    - Insights (velocity, blockers, focus areas)"
)

Velocity Calculation Algorithm

def calculate_velocity(issues_completed, period_days):
    """
    Weighted rolling average velocity.

    Args:
        issues_completed: List of (date, count) tuples
        period_days: Analysis period

    Returns:
        Weighted velocity (issues/day)
    """
    weights = {
        7: 0.50,   # Last week: 50%
        14: 0.30,  # Last 2 weeks: 30%
        30: 0.20,  # Last month: 20%
    }

    total = 0
    for period, weight in weights.items():
        period_issues = sum(c for d, c in issues_completed
                          if d >= now - timedelta(days=period))
        velocity = period_issues / min(period, period_days)
        total += velocity * weight

    return total

Project Status — Kafka Emission (--emit)

The project-status skill supports a --emit flag that serializes a workstream snapshot and relays it to Kafka via the onex-linear-relay CLI. This is the primary ingress path for Linear workstream data into the ONEX event bus.

Usage

# Show dashboard for MVP project + emit snapshot to Kafka
${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/project-status MVP --emit

# All projects overview + emit
${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/project-status --all --emit

# JSON output + emit (useful for scripting)
${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/project-status MVP --json --emit

How Emission Works

1. After generating the status output, --emit serializes a snapshot to /tmp/linear-snapshot-{timestamp}.json:

   {
     "workstreams": ["MVP", "Beta"],
     "project": "MVP - OmniNode Platform Foundation",
     "project_shortcut": "MVP",
     "source": "project-status",
     "generated_at": "2026-02-23T22:00:00+00:00"
   }
   

2. Calls onex-linear-relay emit --snapshot-file /tmp/linear-snapshot-{timestamp}.json
3. The relay publishes to onex.evt.linear.snapshot.v1 (non-blocking, exits 0 even if Kafka is unreachable)

Requirements

• onex-linear-relay must be installed and available on PATH
• Install via: pip install omnibase_infra (or uv add omnibase_infra in your project)
• Related: Phase 2 — effect nodes and CLIs in omnibase_infra

IMPORTANT: Do NOT Call REST Endpoint

The POST /api/linear/snapshot endpoint in omnidash is debug-only ingress. It must never be called from production code or skills. Use --emit (which calls onex-linear-relay) as the only correct production path for Linear data into the event bus.

Skills Location

Claude Code Access: ${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/

Executables:

• ${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/deep-dive - Daily deep dive generator
• ${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/velocity-estimate - Velocity and ETA calculator
• ${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/estimation-accuracy - Three-layer factory telemetry (requires python3)
• ${CLAUDE_PLUGIN_ROOT}/skills/ticketing_insights/project-status - Quick health dashboard with Kafka emission support

See Also

• Linear MCP tools: tracker.*
• Linear ticket skills: ${CLAUDE_PLUGIN_ROOT}/skills/linear/
• PR review skills: ${CLAUDE_PLUGIN_ROOT}/skills/pr-review/
• Deep dive archive: $ONEX_STATE_DIR/deep-dives/
• onex-linear-relay CLI — omnibase_infra package