agestra 4.12.2 → 4.12.4

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "agestra",
-  "version": "4.12.2",
+  "version": "4.12.4",
   "description": "Multi-host AI orchestration toolkit for Claude Code, Codex CLI, and Gemini CLI",
   "type": "module",
   "packageManager": "npm@11.11.0",
@@ -34,7 +34,8 @@
     "test": "turbo test && npm run test:host-assets",
     "test:host-assets": "vitest run scripts/__tests__",
     "dev": "turbo dev",
-    "lint": "turbo lint",
+    "lint": "npm run lint:root && turbo lint",
+    "lint:root": "eslint \"scripts/**/*.mjs\"",
     "clean": "turbo clean",
     "prebundle": "npm run sync:metadata",
     "bundle": "node scripts/bundle.mjs",

package/skills/design.md

@@ -98,6 +98,7 @@ Every generated artifact has a fixed role:
 The JSON consensus ledger is the source of truth. Markdown is a readable artifact only. Do not parse generated Markdown as state and do not hand-edit it to change provider stances, item status, or consensus state.
 
 1. Start an approval-gated structured debate with `agent_debate_structured`.
+   - **mode:** use `"idea"` for exploratory architecture/design option discovery. Use `"review"` only when reviewing an already-written design artifact.
    - **topic:** the design subject.
    - **participants:** only providers reported available by `environment_check` / `provider_list`, plus the host design specialist when the engine supports it.
    - **scope:** the design subject, user constraints, relevant code areas, and any related existing design docs.

package/skills/idea.md

@@ -111,8 +111,8 @@ The `ai_chat` prompt for external providers differs by mode. Use the interview a
       - For each available external provider, call `ai_chat` with `save_as_document`:
         - **save_as_document.kind:** `"individual"`
         - **save_as_document.title:** `Idea Exploration — {provider}`
-        - **save_as_document.metadata:** `{ "Task": "{topic}", "Mode": "Independent", "Round": "0" }`
-        - **prompt:** Use the Mode A or Mode B prompt below.
+        - **save_as_document.metadata:** `{ "Provider": "{provider}", "Task": "{topic}", "Mode": "Independent", "Round": "0" }`
+        - **prompt:** Use the Mode A or Mode B prompt below. The response must include a `<proposals>` block. Each idea is an `<item>` with `id`, `title`, `severity`, and optional `location`; put category, evidence, effort, and priority in the item body.
 
    **Mode A prompt for external providers:**
 
@@ -189,39 +189,28 @@ The `ai_chat` prompt for external providers differs by mode. Use the interview a
         - **title:** `Aggregated Ideas — {topic} — Round 0`
         - **metadata:** `{ "Task": "{topic}", "Mode": "Aggregated", "Round": "0" }`
 
-2. Sequential round-based review (no max — until all agree):
+2. Start the approval-gated structured consensus flow.
 
-   For each round `N` starting at 1:
+   Call `agent_debate_structured` with:
+   - **mode:** `"idea"`.
+   - **topic:** the user's idea-discovery topic.
+   - **participants:** providers reported Available by `environment_check`, plus the host ideator when supported.
+   - **source_documents:** optional; use only when step 1 created individual documents, and pass `{ "document_id": "...", "provider": "..." }` for each document.
+   - **leader:** the current host/leader identity when known.
 
-   a. **Turn order**: external providers alphabetical first, host specialist last when present. This lets the host specialist respond to everyone; more importantly, it lets later providers **read earlier providers' Round N reviews** and either endorse or dissent instead of repeating arguments.
+   The MCP moderator engine extracts proposals, assigns stable `ITEM-*` IDs, sends JSON-only turn packets, validates `{ provider, round, items }` responses, persists `{sessionId}.consensus.json`, and regenerates the debate Markdown from the ledger.
 
-   b. For each provider in order:
-      - Read the working doc (Round 0 aggregated or the revised doc from Round `N-1`).
-      - Read every Round `N` review already written by earlier providers this round (via `workspace_read`).
-      - Send the review prompt including the `<round-votes>` contract described in the `agestra:agestra-moderator` agent's `<Workflow_Document_Review_Round>` section.
-      - Save the reply via `workspace_create_document` with:
-        - **kind:** `"debate"`
-        - **title:** `Round {N} Review — {provider}`
-        - **metadata:** `{ "Task": "{topic}", "Round": "{N}", "Participant": "{provider}" }`
+3. Use the approval gate.
 
-   c. After all providers finish Round `N`, the moderator revises the working doc, marks any unresolved disagreement as Disputed, and decides whether to continue.
+   Poll `agent_debate_status` until `ready-for-approval`, `escalated`, or `error`:
+   - If `ready-for-approval`, call `agent_debate_approve` to write the final synthesis.
+   - If `escalated` or more discussion is needed, call `agent_debate_continue` with 3, 5, or 10 additional rounds.
+   - If the result should be closed without synthesis, call `agent_debate_reject`.
 
-   d. If every `<round-votes>` block for Round `N` has zero `disagree` and the moderator has absorbed every `agree-with-note` → consensus reached. Proceed to step 3.
-
-   e. Otherwise → Round `N+1` with the revised doc.
-
-   f. **Every 10 rounds**: call `AskUserQuestion` to offer `continue`, `stop with current state`, or `escalate split positions`.
-
-3. When consensus (or user-approved stop) is reached, write the final document via `workspace_create_document` with:
-   - **kind:** `"synthesis"`
-   - **title:** `Final — {topic} — Consensus`
-   - **metadata:** `{ "Task": "{topic}", "Mode": "Final Consensus", "Rounds": "{N}", "Participants": "{...}" }`
-   - **content:** Use the `<Final_Consensus_Format>` template from the `agestra:agestra-moderator` agent — per-idea stance matrix, expanded reasoning, clear 동의 현황 summary.
-
-4. Present the final document to the user:
-   - Name the final doc ID and path.
-   - Echo the per-idea stance matrix.
-   - For each Disputed item, show the dissenter's rationale verbatim.
+4. Present the final result to the user:
+   - Name the debate Markdown path, consensus JSON ledger path, and synthesis path if approved.
+   - Echo the per-idea consensus table.
+   - Explain accepted, excluded, and disputed items in plain language.
 
 ### Phase 4: Research (Leader-host-only mode)
 

package/skills/review.md

@@ -75,13 +75,15 @@ Call `environment_check` to determine available providers.
    a. In parallel:
       - Spawn the `agestra:agestra-reviewer` agent for host-local independent analysis.
         After the agent completes, save the host reviewer's result as a document via `workspace_create_document`:
+        - **kind:** `"individual"`
         - **title:** `Code Review — host/reviewer`
-        - **metadata:** `{ "Provider": "host/reviewer", "Task": "{review target}", "Focus": "{selected focus areas}", "Mode": "Independent" }`
+        - **metadata:** `{ "Provider": "host/reviewer", "Task": "{review target}", "Focus": "{selected focus areas}", "Mode": "Independent", "Round": "0" }`
         - **content:** The reviewer agent's full output.
       - For each available provider, call `ai_chat` with `save_as_document`:
+        - **save_as_document.kind:** `"individual"`
         - **save_as_document.title:** `Code Review — {provider}`
-        - **save_as_document.metadata:** `{ "Task": "{review target}", "Focus": "{selected focus areas}", "Mode": "Independent" }`
-        - **prompt:** Review the following code. Focus on: [selected focus areas]. For each finding, provide severity (CRITICAL/HIGH/MEDIUM/LOW), file:line location, and evidence. Target: [the review target]
+        - **save_as_document.metadata:** `{ "Provider": "{provider}", "Task": "{review target}", "Focus": "{selected focus areas}", "Mode": "Independent", "Round": "0" }`
+        - **prompt:** Review the following code. Focus on: [selected focus areas]. For each finding, provide severity (CRITICAL/HIGH/MEDIUM/LOW), file:line location, and evidence. Include a `<proposals>` block where each finding is an `<item>` with `id`, `title`, `severity`, and `location`; put the evidence and impact in the item body. Target: [the review target]
 
    b. Collect all document IDs.
 
@@ -92,9 +94,11 @@ Call `environment_check` to determine available providers.
       - The moderator's integrated document becomes the starting document.
 
 2. Start a structured debate session with `agent_debate_structured`.
-   - Use the integrated document's title/topic as the debate topic.
-   - Reuse the same reviewer set from step 1, excluding `ollama`.
-   - Pass the independent document IDs as source context when available.
+   - **mode:** `"review"`.
+   - **topic:** the review target and selected focus summary.
+   - **participants:** the same reviewer set from step 1, excluding `ollama`.
+   - **source_documents:** optional; use only when step 1 created individual documents, and pass `{ "document_id": "...", "provider": "..." }` for each document.
+   - **leader:** the current host/leader identity when known.
 
 3. Let the MCP moderator engine run the consensus loop.
    - The engine assigns stable `ITEM-*` IDs, builds `{sessionId}.consensus.json`, and sends each provider a JSON turn packet.