RU: Forbid

Add items to the forbidden list during an active loop session. Forbidden items are blocked from opportunity discovery.

Runtime configurable: Works with or without active loop. Changes apply on next iteration.

Self-Evolving Skill: This skill improves through use. If instructions are wrong, parameters drifted, or a workaround was needed — fix this file immediately, don't defer. Only update for real, reproducible issues.

Usage

• /ru:forbid documentation updates - Add "documentation updates" to forbidden list
• /ru:forbid --list - Show current forbidden items
• /ru:forbid --clear - Clear all forbidden items
• /ru:forbid --remove <phrase> - Remove item matching phrase (fuzzy)

Execution

/usr/bin/env bash << 'RALPH_FORBID_SCRIPT'
PROJECT_DIR="${CLAUDE_PROJECT_DIR:-$(pwd)}"
CONFIG_FILE="$PROJECT_DIR/.claude/ru-config.json"

# Get arguments
ARGS="${ARGUMENTS:-}"

# Ensure config file exists with guidance structure
if [[ ! -f "$CONFIG_FILE" ]]; then
    echo '{"guidance": {"forbidden": [], "encouraged": []}}' > "$CONFIG_FILE"
fi

# Ensure guidance structure exists
if ! jq -e '.guidance' "$CONFIG_FILE" >/dev/null 2>&1; then
    jq '. + {guidance: {forbidden: [], encouraged: []}}' "$CONFIG_FILE" > "$CONFIG_FILE.tmp" && mv "$CONFIG_FILE.tmp" "$CONFIG_FILE"
fi

# Handle commands
case "$ARGS" in
    "--list"|"-l")
        echo "Current forbidden items:"
        jq -r '.guidance.forbidden[]?' "$CONFIG_FILE" | while read -r item; do
            echo "  ✗ $item"
        done
        COUNT=$(jq -r '.guidance.forbidden | length' "$CONFIG_FILE")
        echo ""
        echo "Total: $COUNT items"
        ;;
    "--clear"|"-c")
        jq '.guidance.forbidden = []' "$CONFIG_FILE" > "$CONFIG_FILE.tmp" && mv "$CONFIG_FILE.tmp" "$CONFIG_FILE"
        echo "Cleared all forbidden items"
        ;;
    --remove\ *)
        PHRASE="${ARGS#--remove }"
        MATCH=$(jq -r --arg phrase "$PHRASE" \
            '.guidance.forbidden[] | select(. | ascii_downcase | contains($phrase | ascii_downcase))' \
            "$CONFIG_FILE" | head -1)
        if [[ -z "$MATCH" ]]; then
            echo "No forbidden item matches: $PHRASE"
            exit 1
        fi
        TS=$(date -u +%Y-%m-%dT%H:%M:%SZ)
        jq --arg match "$MATCH" --arg ts "$TS" \
            '.guidance.forbidden |= map(select(. != $match)) | .guidance.timestamp = $ts' \
            "$CONFIG_FILE" > "$CONFIG_FILE.tmp" && mv "$CONFIG_FILE.tmp" "$CONFIG_FILE"
        echo "Removed from forbidden list: $MATCH"
        ;;
    "")
        echo "Usage: /ru:forbid <phrase> | --list | --clear | --remove <phrase>"
        echo ""
        echo "Current forbidden items:"
        jq -r '.guidance.forbidden[]?' "$CONFIG_FILE" | while read -r item; do
            echo "  ✗ $item"
        done
        ;;
    *)
        TS=$(date -u +%Y-%m-%dT%H:%M:%SZ)
        jq --arg item "$ARGS" --arg ts "$TS" \
            '.guidance.forbidden = ((.guidance.forbidden // []) + [$item] | unique) | .guidance.timestamp = $ts' \
            "$CONFIG_FILE" > "$CONFIG_FILE.tmp" && mv "$CONFIG_FILE.tmp" "$CONFIG_FILE"
        echo "Added to forbidden list: $ARGS"
        echo ""
        echo "Current forbidden items:"
        jq -r '.guidance.forbidden[]?' "$CONFIG_FILE" | while read -r item; do
            echo "  ✗ $item"
        done
        ;;
esac
RALPH_FORBID_SCRIPT

Run the bash script above to manage forbidden items.

How It Works

1. Config file: Changes are written to .claude/ru-config.json
2. Next iteration applies: The Stop hook reads config fresh on each iteration
3. Template rendering: Forbidden items appear in the ## USER GUIDANCE section

Troubleshooting

Post-Execution Reflection

After this skill completes, check before closing:

1. Did the command succeed? — If not, fix the instruction or error table that caused the failure.
2. Did parameters or output change? — If the underlying tool's interface drifted, update Usage examples and Parameters table to match.
3. Was a workaround needed? — If you had to improvise (different flags, extra steps), update this SKILL.md so the next invocation doesn't need the same workaround.

Only update if the issue is real and reproducible — not speculative.