Stats
Actions
Tags
Help us improve
Share bugs, ideas, or general feedback.
From ui-tokenize
Print the session ledger — blocks, rewrites, fabrications, escalations
npx claudepluginhub xiaolai/ui-tokenize --plugin ui-tokenizeHow this command is triggered — by the user, by Claude, or both
Slash command
/ui-tokenize:metricsThe summary Claude sees in its command listing — used to decide when to auto-load this command
Compact the per-PID NDJSON ledger and print the resulting session metrics. ## Steps 1. **Run the metrics printer.** This compacts the per-PID NDJSON files into the canonical `session.json` first, then reads it. 2. **Surface the report to the user verbatim** using the format below. Do not paraphrase the counts; the user uses these numbers to judge whether the catalog needs more tokens. 3. **Interpret signals.** A high `escapesToProposal` count or repeated fabrications indicate the catalog is missing tokens for the values the agent keeps trying to use; recommend `/tokenize:propose` (...
/metricsDisplays agent performance metrics and session history from .claude/agent-metrics.jsonl: total sessions, recent table, average file changes, trends. Shows 'No metrics recorded yet' if absent.
/metricsDisplays feature-level metrics and analytics dashboard. Supports flags like --recent, --feature, --sprint, --export from /specswarm:metrics.
/metricsAnalyzes team metrics like velocity tracking, cycle time, and performance for a specified metric-type and optional timeframe.
/metricsGenerates quality metrics dashboard report with violation trends by rule, daily sessions, correction patterns from local SQLite database.
/metricsDisplays workflow metrics report on code quality (grade, coverage, linter errors), tests (total/passing, coverage vs target), performance (duration, velocity, efficiency), and workflow (phases, approvals) with pass/warn/fail thresholds. Supports --detailed.
/metricsGenerates real-time metrics dashboard for Jira orchestration activities with SLA compliance, agent performance, quality metrics, and throughput analysis in markdown, ASCII, JSON, or Confluence formats.
Share bugs, ideas, or general feedback.
Compact the per-PID NDJSON ledger and print the resulting session metrics.
node "${CLAUDE_PLUGIN_ROOT}/commands/cli.mjs" metrics
session.json first, then reads it.escapesToProposal count or repeated fabrications indicate the catalog is missing tokens for the values the agent keeps trying to use; recommend /tokenize:propose (or direct tokenize__propose MCP calls) to close the gap.The CLI prints (in order):
Session <session-id> started: <iso-8601> updated: <iso-8601> metrics: <json blob with block / rewrite / escape counts> fabrications: — indented list of token names the agent invented (omitted when empty) unresolved by file: — indented <path>: <count> lines (omitted when empty)If no session ledger exists yet, the CLI prints No session ledger yet. — surface that line directly.
These counts indicate whether the catalog covers the project's needs (high escapesToProposal or repeated fabrications point to catalog gaps).