Agent workflow

Normal observation logging

# Check workspace status
repairledger --json status

# Initialize
repairledger init

# Create via structured bundle
cat <<'JSON' | repairledger repair apply --file - --dry-run
{
  "schema": "repairledger.structured_repair.v1",
  "operation": "create",
  "repair": {
    "title": "Example tool failure",
    "target_tool": "example",
    "kind": "tool_failure",
    "severity": "medium",
    "status": "observed",
    "components": {
      "summary": "Summary here.",
      "context": "Context here.",
      "tool_call": "example command",
      "expected_behavior": "Expected behavior.",
      "actual_behavior": "Actual behavior.",
      "reproduction": "1. Step one.\n2. Step two.",
      "impact": "This matters because...",
      "suggested_fix": "Add validation."
    }
  }
}
JSON

# Create for real
cat <<'JSON' | repairledger repair apply --file -
{ ... same JSON ... }
JSON

# Validate
repairledger repair validate --repair repair-0001 --for-observed

# Build
repairledger repair build --repair repair-0001

# Export report
repairledger report export

Convenience log path

repairledger log \
  --title "Tool rejects valid input" \
  --target-tool mytool \
  --kind validation_false_positive \
  --severity high \
  --summary "Tool rejected valid config." \
  --context "Agent was configuring tool." \
  --tool-call "mytool validate" \
  --expected "Should pass." \
  --actual "Was rejected." \
  --reproduction "1. Run validate. 2. Observe." \
  --impact "Blocks workflow." \
  --suggested-fix "Make field-aware." \
  --status observed