auto-skill telemetry
View usage telemetry and skill effectiveness reports. Auto-Skill collects anonymous metrics to track skill performance and guide graduation decisions.
Usage
auto-skill telemetry [action] [options]
Actions
Show aggregated effectiveness report for all skills
Show raw telemetry events (individual skill uses)
Options
Filter to a specific skill name
Maximum number of events to show (for events action)
Output results in JSON format
What’s Collected
Auto-Skill collects anonymous, aggregate metrics only:
Event Types
Skill loaded, skill used, skill succeeded/failed
Timing Data
Duration in milliseconds for each skill use
Agent IDs
Which coding agent used the skill
OS Platform
Operating system (for compatibility tracking)
NOT collected: PII, search queries, file paths, code content, session transcripts, or any identifiable information.
Privacy Controls
Disable Telemetry
export AUTO_SKILL_NO_TELEMETRY=1
CI Auto-Disable
Telemetry is automatically disabled in CI environments (detects CI=true).
View Source
All telemetry code is open source: src/core/telemetry.ts
Examples
View Effectiveness Report
auto-skill telemetry report
Skill Effectiveness Report
debug-api-errors
Uses: 12 | Success: 92%
Avg Duration: 1847ms
Agents: claude-code, cursor
tdd-workflow
Uses: 8 | Success: 88%
Avg Duration: 3241ms
Agents: claude-code
react-testing-utils
Uses: 6 | Success: 83%
Avg Duration: 2156ms
Agents: claude-code, continue
The effectiveness report shows aggregated data across all uses of each skill, helping you identify which skills are most valuable.
View Raw Events
auto-skill telemetry events
Telemetry Events (latest 20)
[OK] debug-api-errors [claude-code] (1847ms)
Session: a3f2b8c1e4d6... | 2024-03-15T10:23:45.123Z
[OK] tdd-workflow [claude-code] (3241ms)
Session: b9e7c2d1f3a5... | 2024-03-15T09:15:32.456Z
[FAIL] typescript-migration [cursor] (5632ms)
Session: c1d8e3f2a9b6... | 2024-03-15T08:42:18.789Z
[OK] react-testing-utils [continue] (2156ms)
Session: d4f1a7b3c9e2... | 2024-03-14T15:28:51.234Z
Filter to Specific Skill
auto-skill telemetry report --skill debug-api-errors
Skill Effectiveness Report
debug-api-errors
Uses: 12 | Success: 92%
Avg Duration: 1847ms
Agents: claude-code, cursor
auto-skill telemetry events --skill debug-api-errors
Telemetry Events (latest 20)
[OK] debug-api-errors [claude-code] (1847ms)
Session: a3f2b8c1e4d6... | 2024-03-15T10:23:45.123Z
[OK] debug-api-errors [cursor] (1923ms)
Session: e8f5a2b7c3d9... | 2024-03-14T14:15:22.567Z
[OK] debug-api-errors [claude-code] (1756ms)
Session: f9a6b3c8d4e1... | 2024-03-13T11:42:38.891Z
Limit Event Count
auto-skill telemetry events --limit 5
JSON Output - Report
auto-skill telemetry report --json
{
"count": 3,
"reports": [
{
"skillName": "debug-api-errors",
"totalUses": 12,
"successRate": 0.92,
"avgDurationMs": 1847.5,
"agentsUsed": ["claude-code", "cursor"]
},
{
"skillName": "tdd-workflow",
"totalUses": 8,
"successRate": 0.88,
"avgDurationMs": 3241.2,
"agentsUsed": ["claude-code"]
},
{
"skillName": "react-testing-utils",
"totalUses": 6,
"successRate": 0.83,
"avgDurationMs": 2156.7,
"agentsUsed": ["claude-code", "continue"]
}
]
}
JSON Output - Events
auto-skill telemetry events --json
{
"count": 4,
"events": [
{
"skillName": "debug-api-errors",
"agentId": "claude-code",
"sessionId": "a3f2b8c1e4d6f9a2b5c8d1f4e7a9b2c5",
"outcome": "success",
"durationMs": 1847,
"timestamp": "2024-03-15T10:23:45.123Z"
},
{
"skillName": "tdd-workflow",
"agentId": "claude-code",
"sessionId": "b9e7c2d1f3a5b8c4d7e9f1a3b5c7d9e1",
"outcome": "success",
"durationMs": 3241,
"timestamp": "2024-03-15T09:15:32.456Z"
},
{
"skillName": "typescript-migration",
"agentId": "cursor",
"sessionId": "c1d8e3f2a9b6c7d4e1f8a3b5c7d9e1f3",
"outcome": "failure",
"durationMs": 5632,
"timestamp": "2024-03-15T08:42:18.789Z"
},
{
"skillName": "react-testing-utils",
"agentId": "continue",
"sessionId": "d4f1a7b3c9e2d8f5a1b6c3d7e9f2a4b8",
"outcome": "success",
"durationMs": 2156,
"timestamp": "2024-03-14T15:28:51.234Z"
}
]
}
Telemetry Event Structure
Name of the skill that was used
ID of the coding agent that used the skill (e.g., “claude-code”, “cursor”)
Anonymous session identifier (hashed, not traceable)
Result: “success” or “failure”
Duration of skill execution in milliseconds
ISO 8601 timestamp of the event
Effectiveness Report Structure
Total number of times the skill was used
Percentage of successful uses (0.0 - 1.0)
Average duration in milliseconds
List of agent IDs that have used this skill
Use Cases
# Find skills with highest success rate
auto-skill telemetry report --json | jq -r '.reports | sort_by(.successRate) | reverse | .[0:3]'
Track Skill Adoption
# See which agents are using which skills
auto-skill telemetry report --json | jq -r '.reports[] | {skill: .skillName, agents: .agentsUsed}'
Debug Failed Skills
# Find recent failures
auto-skill telemetry events --json | jq -r '.events[] | select(.outcome == "failure")'
Graduation Eligibility
# Check if skill meets graduation criteria (5+ uses, 80%+ success)
auto-skill telemetry report --json | jq -r '.reports[] | select(.totalUses >= 5 and .successRate >= 0.8)'
Telemetry Storage
Telemetry data is stored locally in:
~/.claude/auto-skill/events.db
Telemetry is never sent to external servers. All data stays on your machine.
Interpreting the Data
Success Rate
| Success Rate | Interpretation |
|---|
| 90%+ | Excellent - skill is reliable |
| 80-90% | Good - skill works well |
| 60-80% | Moderate - may need refinement |
| <60% | Poor - skill needs review or removal |
Usage Count
| Uses | Interpretation |
|---|
| 10+ | High adoption - skill is valuable |
| 5-10 | Moderate use - gaining traction |
| 3-5 | Low use - niche or new skill |
| <3 | Rarely used - may not be needed |
Duration
Longer durations may indicate:
- Complex operations (normal for some skills)
- Performance issues
- Network latency (for external skill queries)
Graduate Skills
Use telemetry data to identify graduation candidates
Lock Status
Verify integrity of your most-used skills