From 5ef5a5ac4d7495cdd8f0ae04ff59cc79e77d8aab Mon Sep 17 00:00:00 2001
From: ismeth <ufukaltinok@gmail.com>
Date: Thu, 12 Feb 2026 14:36:56 +0100
Subject: [PATCH] feat(05-02): add confirmation-gated Athena delegation prompt

---
 src/agents/athena/agent.ts | 46 ++++++++++++++++++++++++--------------
 1 file changed, 29 insertions(+), 17 deletions(-)

diff --git a/src/agents/athena/agent.ts b/src/agents/athena/agent.ts
index 28c8239a..cdd0e9c4 100644
--- a/src/agents/athena/agent.ts
+++ b/src/agents/athena/agent.ts
@@ -16,43 +16,55 @@ export const ATHENA_PROMPT_METADATA: AgentPromptMetadata = {
     },
     {
       domain: "Execution planning",
-      trigger: "Need prioritized action targets from council findings without direct delegation",
+      trigger: "Need confirmation-gated delegation after synthesizing council findings",
     },
   ],
   useWhen: [
     "You need Athena to synthesize multi-model council outputs into concrete findings",
     "You need agreement-level confidence before selecting what to execute next",
-    "You need prioritized action targets while keeping final execution control in the caller",
+    "You need explicit user confirmation before delegating fixes to Atlas or planning to Prometheus",
   ],
   avoidWhen: [
     "Single-model questions that do not need council synthesis",
-    "Tasks requiring direct implementation by Athena in this phase",
+    "Tasks requiring direct implementation by Athena",
   ],
 }
 
-const ATHENA_SYSTEM_PROMPT = `You are Athena, a primary synthesis strategist for multi-model council workflows.
+const ATHENA_SYSTEM_PROMPT = `You are Athena, a primary multi-model council strategist. You synthesize independent council member outputs into evidence-grounded findings and delegate execution through confirmation-gated workflows.
 
-Your role in this phase:
-- Synthesize council outputs into evidence-grounded findings.
-- Distinguish consensus from disagreement and flag uncertainty.
-- Recommend prioritized action targets the caller can execute.
+Council Synthesis Workflow:
+1. Receive the user's question and council response set.
+2. Fan out and compare council perspectives as independent evidence inputs.
+3. Collect findings and group them by agreement level: unanimous, majority, minority, solo.
+4. Treat solo findings as potential false positives and call out the risk explicitly.
+5. Present synthesized findings with practical recommendations before any delegation.
+6. Use the established findings presentation pattern from formatFindingsForUser to keep output deterministic and scannable.
 
-Constraints in this phase:
-- Do NOT execute delegation workflows.
-- Do NOT run confirmation-gated delegation logic (reserved for Phase 5).
-- Stay analysis-first and recommendation-oriented.
+Confirmation-Gated Delegation:
+- After presenting findings, ALWAYS wait for explicit user confirmation.
+- NEVER delegate when confirmation is implied, ambiguous, or missing.
+- If the user confirms direct fixes, delegate to Atlas using the task tool with a focused fix prompt.
+- If the user confirms planning, delegate to Prometheus using the task tool with a focused plan prompt.
+- Build delegation prompts using the established delegation prompt patterns.
+- Include both the original question and only the confirmed findings in the delegation prompt context.
 
-Output requirements:
-- Keep conclusions concrete, reference supporting evidence from council responses, and call out confidence level.
-- Provide clear next action targets ordered by impact and confidence.
-- Explicitly identify disagreement areas and false-positive risk.`
+Output Format:
+- Present findings grouped by agreement level in this order: unanimous, majority, minority, solo.
+- For each finding, include Athena's assessment and rationale.
+- End with clear action options: "fix now" (Atlas) or "create plan" (Prometheus).
+- Ask the user to confirm which findings to act on and which action path to take.
+
+Constraints:
+- Do NOT write or edit files directly.
+- Do NOT delegate without explicit user confirmation.
+- Do NOT ignore solo finding false-positive warnings.`
 
 export function createAthenaAgent(model: string): AgentConfig {
   const restrictions = createAgentToolRestrictions(["write", "edit"])
 
   const base = {
     description:
-      "Primary synthesis strategist for multi-model council outputs. Produces evidence-grounded findings and prioritized action targets without executing delegation workflows. (Athena - OhMyOpenCode)",
+      "Primary synthesis strategist for multi-model council outputs. Produces evidence-grounded findings and runs confirmation-gated delegation to Atlas (fix) or Prometheus (plan) via task tool. (Athena - OhMyOpenCode)",
     mode: MODE,
     model,
     temperature: 0.1,