JSON Parse (16)

JSON Parse (16) is exactly what it sounds like: the pack's JSON Parse node, but with sixteen path fields and sixteen out ports instead of eight. Same dot/bracket path syntax, same code-fence stripping, same default fallback, same obj output carrying the whole parsed object. If you've read the eight-port version, you've read this node; the only real difference is the headroom.

When you actually need sixteen

A typical LLM JSON has a handful of fields. Eight ports usually covers it - which is why the eight-port node is the pack's default. You reach for the 16-port variant when your schema genuinely sprawls: a rich character card with appearance, personality, speech style, relationships, plus image params; a big settings dump from a config-generating LLM; a multi-stage prompt where you want every segment pre-split. It's also handy when you want to route a JSON through and keep spare ports for fields you'll add later without rewiring.

The inputs and outputs that matter

• json_text - the JSON string (fence-stripping on by default via strip_code_fence).
• default - what any missing path resolves to.
• path1…path16 - the extraction paths. Empty paths just leave their output at the default.
• Outputs out1…out16 (all STRING) plus obj (SIMPLECHAT_JSON).

Paths work like a.b.c, items[0].name, or items[0]["name"]. Values stringify like the 8-port node: numbers become text, nested objects become JSON text. Nothing about the semantics changes - it's a strict json.loads under the hood, so feed it clean JSON or it errors.

Installing it

cd ComfyUI/custom_nodes
git clone https://github.com/Moeblack/ComfyUI-SimpleChat

Restart, or use Manager and search "ComfyUI-SimpleChat". Only dependency: aiohttp.

Gotchas

Sixteen ports of pure string output is a lot of manual wiring, and every value still lands as STRING - no INT/FLOAT conversion, so numbers heading into numeric inputs still need casting. If most of your paths are empty you're carrying dead weight; the 8-port version is tidier for small schemas, and for the fixed Anima prompt schema Prompt JSON Unpack beats both because it emits typed outputs and a vars dict in one go. Only pay the 16-port sprawl when the schema actually demands it.