Sign in to continue
You need to sign in with your organisation account to use the Guideline AI Agent.
Sent to the LLM for Mode A (definitional) questions. No clarifying questions — answer directly from policy.
Sent to the LLM for Mode B (program-known) questions. May ask one clarifying question if a dimension genuinely splits the answer.
Sent to the LLM for Mode C (scenario) questions. Identifies candidate programs, presents comparisons, at most one follow-up.
The original single system prompt. Used for approve/reset baseline management. Changes here do not affect Mode A/B/C routing.
Drives the Phase 1 analysis LLM for Mode A questions. Must never set needs_clarification: true.
Drives the Phase 1 analysis LLM for Mode B questions. Clarifies only when a dimension genuinely splits the answer.
Drives the Phase 1 analysis LLM for Mode C questions. Identifies candidate programs, one follow-up max.
The original single analysis prompt. Kept for reference.
This text appears at the end of every AI response. Changes take effect within 5 minutes.
Every prompt save and reset, newest first. Showing last — of — entries.