{# CUI // SP-CTI β€” Reasoned Codegen panel (tools/llm/reasoned_codegen*.py) #} {% set rc = data.reasoned_codegen if data is defined and data.reasoned_codegen is defined else {} %}

🧠 Reasoned Codegen {% if rc.section_enabled is sameas false %} KILL-SWITCH OFF {% else %} ENABLED {% endif %}

CoT/CoD generation + adversarial critique + verify/repair for code pipelines. Section kill-switch + per-function defaults from args/llm_config.yaml.

{# ── Per-function config ────────────────────────────────────────────── #} {% for fn in rc.get('functions', []) %} {% else %} {% endfor %}
FunctionEnabledMode CritiqueRepair rounds
{{ fn.function }} {{ 'on' if fn.enabled else 'off' }} {{ fn.mode }} {{ 'βœ“' if fn.critique else 'β€”' }} {{ fn.max_repair_rounds }}
No per-function config found.
Caps: cost ≀ ${{ rc.get('cost_cap_usd', 'β€”') }} Β· tokens ≀ {{ rc.get('token_cap', 'β€”') }} Β· default mode {{ rc.get('default_mode', 'off') }}
{# ── Advisor "try it" ───────────────────────────────────────────────── #}

Advisor β€” should this task use reasoned codegen?

{# ── Recent chain runs ──────────────────────────────────────────────── #}

Recent reasoned / chain runs

Loading…