tunelisted

Analyze evaluation logs and maintain the project knowledge base. User arguments: $ARGUMENTS ## Pre-computed Context Active tickets: !`ls -d .simple-workflow/backlog/active/*/ 2>/dev/null | head -10` Done tickets: !`ls -d .simple-workflow/backlog/done/*/ 2>/dev/null | head -10` Knowledge base status: !`ls .simple-workflow/kb/*.yaml 2>/dev/null || echo "KB not initialized"` Available user skills: !`( ls -1 ~/.claude/skills 2>/dev/null ; ls -1 .claude/skills 2>/dev/null ) | sort -u | grep . | tr "\n" "," | sed "s/,$//" | grep . || echo "(none)"` ## YAML Schema Reference ### entries.yaml ```yaml entries: - id: "entry-001" pattern: "description" category: "error-handling|security|performance|convention|testing|decision" scope: "bash|typescript|python|general" roles: ["implementer"] confidence: 0.92 evidence_count: 4 sources: - ticket: "slug" round: 1 type: "eval-round" observed_at: "2026-04-01" promoted_at: "2026-04-10" ``` ### index.yaml ```yaml implementer: - id: "entry-001" summary: "one-line summary" confidence: 0.92 ``` ### candidates.yaml ```yaml candidates: - id: "cand-001" pattern: "description" category: "category" scope: "scope" roles: ["role1"] confidence: 0.3 evidence_count: 1 sources: - ticket: "slug" round: 1 type: "eval-round" observed_at: "2026-04-10" ``` Invocation policy: Do not auto-invoke. Only invoke when explicitly