# Metrics > >- - **Type:** Skill - **Install:** `agentstack add skill-samibs-skillfoundry-metrics` - **Verified:** Yes — security-reviewed for prompt injection and unsafe behavior - **Seller:** [samibs](https://agentstack.voostack.com/s/samibs) - **Installs:** 0 - **Category:** [AI & ML](https://agentstack.voostack.com/c/ai-and-ml) - **Latest version:** 0.1.0 - **License:** MIT - **Upstream author:** [samibs](https://github.com/samibs) - **Source:** https://github.com/samibs/skillfoundry/tree/main/.agents/skills/metrics - **Website:** https://skillfoundry.work ## Install ```sh agentstack add skill-samibs-skillfoundry-metrics ``` Requires the [AgentStack CLI](https://agentstack.voostack.com/docs/cli). Works with Claude Code, Cursor, and any MCP-compatible agent. ## About # Metrics Dashboard View and analyze execution metrics for the Claude AS framework. --- ## OPERATING MODE ``` /metrics → Show metrics dashboard /metrics agents → Agent performance breakdown /metrics stories → Story completion analysis /metrics errors → Error analysis and patterns /metrics trends → Trend analysis over time /metrics export [format] → Export metrics (json/csv/md) /metrics reset → Reset metrics (with confirmation) ``` --- ## DASHBOARD VIEW On `/metrics` invocation, display the comprehensive dashboard: ``` METRICS DASHBOARD ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ PROJECT: [project-name] PERIOD: All time (since [start-date]) QUICK STATS ┌────────────────────────────────────────────────────────────────┐ │ Executions: [N] │ Success: [X]% │ Stories: [N] │ Tokens: [X]K │ └────────────────────────────────────────────────────────────────┘ RECENT ACTIVITY (Last 7 days) ├── Executions: [N] ├── Stories Completed: [N] ├── Success Rate: [X]% └── Avg Tokens/Execution: [X]K TOP PERFORMING AGENTS 1. [agent-name]: [X]% success ([N] calls) 2. [agent-name]: [X]% success ([N] calls) 3. [agent-name]: [X]% success ([N] calls) ATTENTION NEEDED [List any metrics below threshold] Use /metrics [category] for detailed breakdowns. ``` --- ## AGENT METRICS On `/metrics agents`: ``` AGENT PERFORMANCE METRICS ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ OVERVIEW ┌──────────────────┬──────────┬──────────┬──────────┬──────────┬──────────┐ │ Agent │ Calls │ Success │ Partial │ Failed │ Avg Tok │ ├──────────────────┼──────────┼──────────┼──────────┼──────────┼──────────┤ │ ruthless-coder │ [N] │ [X]% │ [X]% │ [X]% │ [X]K │ │ ruthless-tester │ [N] │ [X]% │ [X]% │ [X]% │ [X]K │ │ gate-keeper │ [N] │ [X]% │ [X]% │ [X]% │ [X]K │ │ cold-architect │ [N] │ [X]% │ [X]% │ [X]% │ [X]K │ │ docs-codifier │ [N] │ [X]% │ [X]% │ [X]% │ [X]K │ │ debug-hunter │ [N] │ [X]% │ [X]% │ [X]% │ [X]K │ └──────────────────┴──────────┴──────────┴──────────┴──────────┴──────────┘ PERFORMANCE RANKING 1. 🥇 [agent]: [X]% success rate 2. 🥈 [agent]: [X]% success rate 3. 🥉 [agent]: [X]% success rate TOKEN EFFICIENCY Most Efficient: [agent] ([X] tokens/call avg) Least Efficient: [agent] ([X] tokens/call avg) RECENT FAILURES BY AGENT ┌──────────────────┬──────────┬────────────────────────────────────┐ │ Agent │ Failures │ Common Cause │ ├──────────────────┼──────────┼────────────────────────────────────┤ │ [agent] │ [N] │ [most common error type] │ └──────────────────┴──────────┴────────────────────────────────────┘ RECOMMENDATIONS [Based on agent performance, suggest improvements] ``` --- ## STORY METRICS On `/metrics stories`: ``` STORY COMPLETION METRICS ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ OVERALL ├── Total Stories: [N] ├── Completed: [N] ([X]%) ├── Failed: [N] ([X]%) ├── Blocked: [N] ([X]%) └── Avg Tokens/Story: [X]K BY COMPLEXITY ┌────────────┬───────┬──────────┬──────────┬──────────┐ │ Complexity │ Count │ Success │ Failed │ Avg Tok │ ├────────────┼───────┼──────────┼──────────┼──────────┤ │ Simple │ [N] │ [X]% │ [X]% │ [X]K │ │ Medium │ [N] │ [X]% │ [X]% │ [X]K │ │ Complex │ [N] │ [X]% │ [X]% │ [X]K │ └────────────┴───────┴──────────┴──────────┴──────────┘ BY LAYER ┌────────────┬───────┬──────────┬──────────┐ │ Layer │ Count │ Success │ Trend │ ├────────────┼───────┼──────────┼──────────┤ │ Database │ [N] │ [X]% │ [trend] │ │ Backend │ [N] │ [X]% │ [trend] │ │ Frontend │ [N] │ [X]% │ [trend] │ │ Tests │ [N] │ [X]% │ [trend] │ │ Docs │ [N] │ [X]% │ [trend] │ └────────────┴───────┴──────────┴──────────┘ MOST COMMON FAILURE POINTS 1. [failure type]: [N] occurrences 2. [failure type]: [N] occurrences 3. [failure type]: [N] occurrences RECOMMENDATIONS [Based on story patterns, suggest improvements] ``` --- ## ERROR METRICS On `/metrics errors`: ``` ERROR ANALYSIS ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ SUMMARY ├── Total Errors: [N] ├── Recovery Rate: [X]% ├── Rollbacks Triggered: [N] └── Avg Errors/Execution: [X] BY ERROR TYPE ┌────────────────────┬───────┬──────────┬──────────┐ │ Error Type │ Count │ % Total │ Recovered│ ├────────────────────┼───────┼──────────┼──────────┤ │ TEST_FAILED │ [N] │ [X]% │ [X]% │ │ BUILD_FAILED │ [N] │ [X]% │ [X]% │ │ BLOCKED │ [N] │ [X]% │ [X]% │ │ TIMEOUT │ [N] │ [X]% │ [X]% │ │ INVALID_INPUT │ [N] │ [X]% │ [X]% │ │ SECURITY_VIOLATION │ [N] │ [X]% │ [X]% │ └────────────────────┴───────┴──────────┴──────────┘ BY AGENT ┌──────────────────┬───────┬──────────────────────────────────┐ │ Agent │ Errors│ Most Common │ ├──────────────────┼───────┼──────────────────────────────────┤ │ [agent] │ [N] │ [error type] │ │ [agent] │ [N] │ [error type] │ └──────────────────┴───────┴──────────────────────────────────┘ RECENT ERRORS (Last 10) ┌────────────────┬────────────────┬────────────┬──────────┐ │ Time │ Story │ Error │ Resolved │ ├────────────────┼────────────────┼────────────┼──────────┤ │ [timestamp] │ STORY-XXX │ [type] │ [yes/no] │ └────────────────┴────────────────┴────────────┴──────────┘ ROOT CAUSE ANALYSIS [Pattern identification and recommendations] ``` --- ## TREND ANALYSIS On `/metrics trends`: ``` TREND ANALYSIS ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ PERIOD: Last 30 days EXECUTION TREND Week 1: ████████░░ 8 executions Week 2: ██████████ 10 executions Week 3: ████████████ 12 executions Week 4: ███████████████ 15 executions Trend: ↑ +87.5% growth SUCCESS RATE TREND Week 1: 91.2% ░░░░░░░░░░ Week 2: 92.5% ████░░░░░░ Week 3: 94.1% ████████░░ Week 4: 95.2% ██████████ Trend: ↑ +4.0% improvement TOKEN EFFICIENCY TREND Week 1: 52K tokens/exec Week 2: 48K tokens/exec Week 3: 45K tokens/exec Week 4: 42K tokens/exec Trend: ↓ -19.2% (improving) WEEKLY COMPARISON ┌────────────────┬──────────┬──────────┬──────────┐ │ Metric │ This Wk │ Last Wk │ Change │ ├────────────────┼──────────┼──────────┼──────────┤ │ Executions │ [N] │ [N] │ [+/-X%] │ │ Success Rate │ [X]% │ [X]% │ [+/-X%] │ │ Stories │ [N] │ [N] │ [+/-X%] │ │ Errors │ [N] │ [N] │ [+/-X%] │ │ Avg Tokens │ [X]K │ [X]K │ [+/-X%] │ └────────────────┴──────────┴──────────┴──────────┘ PROJECTIONS (Next 7 days) Based on current trends: ├── Expected Executions: [N] ├── Expected Success Rate: [X]% └── Expected Token Usage: [X]K HEALTH INDICATORS ├── Execution Volume: [HEALTHY/WARNING/CRITICAL] ├── Success Rate: [HEALTHY/WARNING/CRITICAL] ├── Token Efficiency: [HEALTHY/WARNING/CRITICAL] └── Error Rate: [HEALTHY/WARNING/CRITICAL] ``` --- ## EXPORT On `/metrics export [format]`: ### JSON Export ```bash /metrics export json Exporting metrics to: .claude/metrics/export_20260120.json ├── Summary data ├── Agent metrics ├── Story metrics ├── Error metrics └── Trend data Export complete: .claude/metrics/export_20260120.json ``` ### CSV Export ```bash /metrics export csv Exporting metrics to: .claude/metrics/export_20260120/ ├── summary.csv ├── agents.csv ├── stories.csv ├── errors.csv └── daily.csv Export complete: .claude/metrics/export_20260120/ ``` ### Markdown Report ```bash /metrics export md Generating report: .claude/metrics/report_20260120.md ├── Executive summary ├── Detailed metrics ├── Charts (ASCII) ├── Recommendations Report generated: .claude/metrics/report_20260120.md ``` --- ## RESET On `/metrics reset`: ``` ⚠️ RESET METRICS This will permanently delete all collected metrics. Current metrics summary: ├── Executions: [N] ├── Stories: [N] ├── Days of data: [N] └── Size: [X] KB Type 'RESET' to confirm, or 'cancel' to abort: ``` After confirmation: ``` Metrics reset complete. Backed up to: .claude/metrics/backup_20260120.json New metrics file initialized. ``` --- ## METRICS FILE LOCATION ``` .claude/ ├── metrics.json # Current metrics ├── metrics/ │ ├── summary.json # Aggregated summary │ ├── daily/ │ │ └── YYYY-MM-DD.json # Daily snapshots │ ├── executions/ │ │ └── exec_*.json # Per-execution records │ ├── exports/ │ │ └── export_*.json # Exported data │ └── backups/ │ └── backup_*.json # Reset backups ``` --- ## THRESHOLDS Default thresholds for health indicators: | Metric | Healthy | Warning | Critical | |--------|---------|---------|----------| | Success Rate | > 95% | 85-95% | 90% | 80-90% | 15% | | Recovery Rate | > 90% | 75-90% | 15% | CRITICAL | "Error rate critically high. Review root causes." | | Recovery rate 20% week-over-week | WARNING | "Token usage increasing. Review efficiency." | | 0 executions in 7+ days | INFO | "No recent activity. Metrics may be stale." | ### Threshold Configuration Thresholds can be overridden in `.claude/metrics-config.json`: ```json { "thresholds": { "success_rate": { "healthy": 95, "warning": 85, "critical": 75 }, "agent_success": { "healthy": 90, "warning": 80, "critical": 70 }, "error_rate": { "healthy": 5, "warning": 15, "critical": 25 }, "recovery_rate": { "healthy": 90, "warning": 75, "critical": 60 } } } ``` --- ## BAD vs GOOD Examples ### BAD: Metrics dashboard with no context or actionability ``` /metrics METRICS DASHBOARD ━━━━━━━━━━━━━━━━━━ Executions: 47 Success: 89% Stories: 142 Tokens: 2.1M ``` Problem: Raw numbers without context. No trend direction. No comparison to thresholds. No recommendations. User cannot tell if 89% is good or bad, or whether things are improving or degrading. ### GOOD: Metrics dashboard with trends, alerts, and recommendations ``` /metrics METRICS DASHBOARD ━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━ PROJECT: skillfoundry PERIOD: All time (since 2026-01-15) QUICK STATS ┌────────────────────────────────────────────────────────────────┐ │ Executions: 47 │ Success: 93.6% │ Stories: 142 │ Tokens: 2.1M │ └────────────────────────────────────────────────────────────────┘ RECENT ACTIVITY (Last 7 days) ├── Executions: 12 (↑ +20% vs prior week) ├── Stories Completed: 38 ├── Success Rate: 95.2% (↑ +1.6% — HEALTHY) └── Avg Tokens/Execution: 42K (↓ -8% — improving) TOP PERFORMING AGENTS 1. ruthless-coder: 97% success (34 calls) 2. ruthless-tester: 94% success (34 calls) 3. cold-architect: 92% success (34 calls) ATTENTION NEEDED ⚠ gate-keeper: 78% success rate (below 80% threshold) → Most common failure: incomplete security checklist → Recommendation: Add security section to story template Use /metrics [category] for detailed breakdowns. ``` --- ## REFLECTION PROTOCOL ### Pre-Execution Reflection Before generating metrics, answer: - Is the metrics data file (`metrics.json`) present and non-empty? - What time period is the user interested in? - Are there enough data points for meaningful trends (need 7+ days)? - Should I highlight any known data quality issues? ### Post-Execution Reflection After displaying metrics, evaluate: - Did I provide context for every number (trend, threshold comparison)? - Were recommendations specific and actionable? - Did I identify the most important metric to focus on? - Were calculations verified (no division by zero, no stale data)? ### Self-Score (1-10) | Dimension | Score | Criteria | |-----------|-------|----------| | Accuracy | [1-10] | Were all calculations correct? No math errors? | | Actionability | [1-10] | Can the user take specific action based on this dashboard? | | Completeness | [1-10] | Were all relevant metrics categories displayed? | | Clarity | [1-10] | Is the dashboard easy to read and understand? | **Threshold**: If Accuracy scores below 8, re-verify calculations before displaying. If Actionability scores below 6, add specific recommendations. --- ## ERROR HANDLING | Error | Cause | Resolution | |-------|-------|------------| | Metrics file not found | No executions yet or file deleted | Display "No metrics collected yet. Run /go to start." | | Metrics file corrupted | Malformed JSON | Attempt recovery from daily backups; if none, reset | | Insufficient data for trends | Fewer than 7 days of data | Show raw numbers only, note trends unavailable | | Division by zero | No executions in period | Display "N/A" for rates, note no activity | | Export fails | Disk full or permission error | Report error, suggest checking disk space | | Stale data warning | No new data in 7+ days | Note data may not reflect current state | --- ## INTEGRATION WITH OTHER AGENTS | Agent | Interaction | |-------|------------| | `/go` | Primary data source — records execution lifecycle metrics | | `/forge` | Records forge pipeline metrics (per-phase timing) | | `/anvil` | Reports per-tier pass/fail rates | | `/fixer` | Reports auto-fix success rates and retry counts | | `/status` | Includes key metrics in project status dashboard | | `/health` | Uses metric thresholds as health indicators | | `/cost` | Token usage metrics feed into cost analysis | | `/analytics` | Deeper agent-level analytics complement metrics | ### Peer Improvement Signals - **From `/go`**: Execution completion events trigger metric recording - **From `/anvil`**: Per-tier pass/fail data enriches quality metrics - **From `/fixer`**: Auto-fix success rates indicate remediation effectiveness - **To `/health`**: Provide health indicator data (success rate, error rate) - **To `/status`**: Provide summary KPIs for project dashboard - **To `/cost`**: Provide token usage data for cost projections --- ## REMEMBER > "Metrics are the mirror of your process. Look honestly." > "Every number tells a story. Learn to read it." > "Good metrics lead to good decisions." ## Source & license This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code. - **Author:** [samibs](https://github.com/samibs) - **Source:** [samibs/skillfoundry](https://github.com/samibs/skillfoundry) - **License:** MIT - **Homepage:** https://skillfoundry.work Install and usage instructions live in the source repository linked above. ## Pricing - **Free** — Free ## Security capabilities Automated source analysis of v0.1.0 — what this tool can access: - **Network access:** no - **Filesystem access:** no - **Shell / process execution:** no - **Environment & secrets:** no - **Dynamic code execution:** no *"Yes" means the capability is present in the source — more access means more to trust, not that it is unsafe.* ## Versions - **0.1.0** — security scan: passed — Imported from the upstream source. ## Links - Listing page: https://agentstack.voostack.com/l/skill-samibs-skillfoundry-metrics - Seller: https://agentstack.voostack.com/s/samibs - Browse the marketplace: https://agentstack.voostack.com/browse --- Listed on AgentStack — the marketplace for AI agent skills and MCP servers. Every listing is security-reviewed. Creators keep 70%.