reduce

func CleanToolOutputJSON ¶

func CleanToolOutputJSON(content string) string

CleanToolOutputJSON extracts plain text from Tool Output turns whose body is raw JSON from the Claude API (e.g. [{"type":"text","text":"..."}]). The JSON text field's \n escapes become real newlines.

Parameters:

• content: Journal entry content with potential JSON tool output

Returns:

• string: Content with JSON tool output replaced by plain text

func StripFences ¶

func StripFences(content string, fencesVerified bool) string

StripFences removes all code fence markers from content, leaving the inner text as-is. This eliminates fence nesting conflicts entirely. Files whose fences have been verified (fencesVerified=true) are returned unchanged.

The result is plain text with structural markers preserved (turn headers, tool calls, section breaks). Serves as a readable baseline without AI reconstruction, or as input for the ctx-journal-normalize skill.

Parameters:

• content: Raw Markdown content of a journal entry
• fencesVerified: Whether the file's fences have been verified via state

Returns:

• string: Content with code fence markers removed

func StripSystemReminders ¶

func StripSystemReminders(content string) string

StripSystemReminders removes internal Claude Code blocks from journal content. Handles:

• XML-style system reminders: <system-reminder>...</system-reminder>
• Bold-style system reminders: **System Reminder**: ... (paragraph until blank line)
• Context compaction summaries: multi-line <summary>...</summary> blocks (standalone <summary> on its own line - see config.TagCompactionSummaryOpen)
• Compaction continuation boilerplate: "If you need specific details from before compaction..." paragraph

The authoritative JSONL transcripts retain them; the exported Markdown doesn't need them.

Parameters:

• content: Journal entry content with potential internal blocks

Returns:

• string: Content with all internal blocks removed