$ cat node-template.py
Human Codex Analysis
// Analysis half of the Human Codex Interview as a headless macro: takes an already-typed interview transcript and the report settings, runs the four-pass Human Codex engine with the built-in Knowledge Base v2 (48 dimensions, 398 signed causal relations) and returns the schema v2.0 profile JSON, the second-person Markdown report, the headline archetype and validation notes. Since 1.1.0 it also stores the result in the organization ontology (Apollo48EvaluatedUser + Apollo48Evaluation, keyed by the `evaluation` input, linked by has_evaluation) through Ontology Create Object / Create Link nodes with the caller's token, and returns the object ids; a failed write fails the run. No interview loop, no Drive writes.
1import json2import sys345def main():6 print(json.dumps({7 "error": "Macro nodes are executed by the macro executor, not directly. "8 "Please configure the macro using the wizard.",9 }), file=sys.stderr)10 sys.exit(1)111213if __name__ == "__main__":14 main()15$ git log --oneline
1.1.0: new required input `evaluation` (JSON identity of the run) and new outputs ontology_evaluation_id, ontology_user_id and ontology_link_id; a Build Ontology Records node (no writes) feeds two Ontology Create Object nodes (Apollo48EvaluatedUser, Apollo48Evaluation with the full result) and an Ontology Create Link node (has_evaluation), which write with the caller's token. A failed write fails the run. 1.0.0: cut from the Human Codex Interview workspace template 1.1.0 (Pass 1-4 + Assemble, Knowledge Base v2).
1.0.2: restore the curated input/output schemas and tags that the 1.0.1 editor republish replaced (input version_hash back to 83bf7b29, which pinned callers send). Keeps 1.0.1's Qwen 3.6 pins on Pass 2-4.
Default Qwen LLM 3.6
1.0.0: cut from the Human Codex Interview workspace template 1.1.0 (Pass 1-4 + Assemble, Knowledge Base v2).