diff --git a/Packs/pai-council-skill/README.md b/Packs/pai-council-skill/README.md
index ed8e1dd06..9c22b87be 100755
--- a/Packs/pai-council-skill/README.md
+++ b/Packs/pai-council-skill/README.md
@@ -1,14 +1,14 @@
 ---
 name: PAI Council Skill
-pack-id: danielmiessler-pai-council-skill-v2.3.0
-version: 2.3.0
+pack-id: danielmiessler-pai-council-skill-v2.5.0
+version: 2.5.0
 author: danielmiessler
-description: Multi-agent debate system where specialized agents discuss topics in rounds, respond to each other's points, and surface insights through intellectual friction.
+description: Multi-agent debate system where specialized agents discuss topics in rounds, respond to each other's points, and surface insights through intellectual friction. Now with adaptive rounds, file-first output, and recovery mode.
 type: skill
-purpose-type: [decision-making, collaboration, multi-agent, debate, synthesis]
+purpose-type: [decision-making, collaboration, multi-agent, debate, synthesis, recovery, resilience]
 platform: claude-code
 dependencies: [pai-core-install]
-keywords: [council, debate, multi-agent, perspectives, synthesis, decision, architect, designer, engineer, researcher]
+keywords: [council, debate, multi-agent, perspectives, synthesis, decision, architect, designer, engineer, researcher, recovery, adaptive, patchlist]
 ---
 
 <p align="center">
@@ -25,11 +25,15 @@ keywords: [council, debate, multi-agent, perspectives, synthesis, decision, arch
 
 ## What This Pack Provides
 
-- **Multi-Agent Debate System** - Structured 3-round debates between specialized agents
+- **Multi-Agent Debate System** - Structured 2-3 round debates between specialized agents
+- **Adaptive Rounds** - Round 3 runs only when needed based on convergence detection
+- **File-First Output** - Session state persists to `~/.claude/MEMORY/` for resilience
+- **Recovery Mode** - Resume interrupted sessions from checkpoint
 - **Quick Consensus Check** - Fast single-round perspective gathering
 - **Council Members** - Architect, Designer, Engineer, Researcher (with optional Security, Intern, Writer)
 - **Visible Transcripts** - Full conversation history showing intellectual friction
-- **Synthesis Engine** - Convergence detection and recommendation generation
+- **Output Modes** - Deliberative (conversational) or Patchlist (structured changes)
+- **Model Tiering** - Automatic model selection per round for cost efficiency
 
 ## Key Differentiator
 
@@ -44,20 +48,23 @@ keywords: [council, debate, multi-agent, perspectives, synthesis, decision, arch
 ```
 Council Skill
 ├── SKILL.md              # Main entry point and routing
+├── Config.md             # Default settings, model tiering, output modes
 ├── CouncilMembers.md     # Agent roles, perspectives, voices
-├── RoundStructure.md     # 3-round debate structure
+├── RoundStructure.md     # Round structure and timing
 ├── OutputFormat.md       # Transcript format templates
 └── Workflows/
-    ├── Debate.md         # Full 3-round structured debate
-    └── Quick.md          # Fast single-round check
+    ├── Debate.md         # Full 2-3 round structured debate (adaptive)
+    ├── Quick.md          # Fast single-round check
+    └── Recovery.md       # Resume interrupted sessions
 ```
 
 ## Workflows
 
 | Workflow | Purpose | Rounds | Output |
 |----------|---------|--------|--------|
-| **DEBATE** | Full structured discussion | 3 | Complete transcript + synthesis |
+| **DEBATE** | Full structured discussion | 2-3 (adaptive) | Complete transcript + synthesis |
 | **QUICK** | Fast perspective check | 1 | Initial positions only |
+| **RECOVERY** | Resume interrupted session | Remaining | Continue from checkpoint |
 
 ## Default Council Members
 
@@ -68,25 +75,33 @@ Council Skill
 | **Engineer** | Implementation reality, tech debt | Marcus Webb |
 | **Researcher** | Data, precedent, external examples | Ava Chen |
 
-## Three-Round Debate Structure
+## Adaptive Round Debate Structure
 
-1. **Round 1 - Initial Positions**: Each agent states their perspective
-2. **Round 2 - Responses & Challenges**: Agents respond to each other's points
-3. **Round 3 - Synthesis**: Identify convergence, disagreements, final recommendations
+1. **Round 1 - Initial Positions**: Each agent states their perspective (sonnet)
+2. **Round 2 - Responses & Challenges**: Agents respond to each other's points (haiku)
+3. **Round 3 - Synthesis** (conditional): Only runs if low convergence or blocking issues (sonnet)
 
-**Total Time:** 30-90 seconds for full debate (parallel execution within rounds)
+**Convergence Check:** After Round 2, the orchestrator evaluates whether agents have reached consensus. Round 3 is skipped if 3+ agents agree and no blocking issues remain.
+
+**Total Time:** 20-60 seconds (2 rounds) or 30-90 seconds (3 rounds)
 
 ## Usage Examples
 
 ```
 "Council: Should we use WebSockets or SSE?"
--> Invokes DEBATE workflow -> 3-round transcript
+-> Invokes DEBATE workflow -> 2-3 round transcript
 
 "Quick council check: Is this API design reasonable?"
 -> Invokes QUICK workflow -> Fast perspectives
 
 "Council with security: Evaluate this auth approach"
 -> DEBATE with Security agent added
+
+"Council (patchlist): Review these specifications"
+-> DEBATE with structured output format
+
+"Council recovery: Resume session 20260202-143052-a1b2c3d4"
+-> RECOVERY workflow -> Continue from checkpoint
 ```
 
 ## What's Included
@@ -94,11 +109,13 @@ Council Skill
 | Component | File | Purpose |
 |-----------|------|---------|
 | Main Skill | src/skills/Council/SKILL.md | Entry point and workflow routing |
+| Configuration | src/skills/Council/Config.md | Default settings, model tiering, output modes |
 | Council Members | src/skills/Council/CouncilMembers.md | Agent roles and voice mapping |
 | Round Structure | src/skills/Council/RoundStructure.md | Debate timing and phases |
 | Output Format | src/skills/Council/OutputFormat.md | Transcript templates |
-| Debate Workflow | src/skills/Council/Workflows/Debate.md | Full 3-round debate |
+| Debate Workflow | src/skills/Council/Workflows/Debate.md | Full 2-3 round adaptive debate |
 | Quick Workflow | src/skills/Council/Workflows/Quick.md | Fast consensus check |
+| Recovery Workflow | src/skills/Council/Workflows/Recovery.md | Resume interrupted sessions |
 
 ## Integration
 
@@ -115,6 +132,15 @@ Council Skill
 
 ## Changelog
 
+### 2.5.0 - 2026-02-02
+- **Adaptive Rounds** - Round 3 now conditional based on convergence detection
+- **File-First Output** - Session state persists to `~/.claude/MEMORY/` for resilience against context compaction and rate limits
+- **Recovery Workflow** - Resume interrupted sessions with partial or full rerun options
+- **Config.md** - Centralized configuration for rounds, agents, output modes, and model tiering
+- **Output Modes** - Added `patchlist` mode for structured specification reviews
+- **Model Tiering** - Automatic model selection (sonnet/haiku) per round for cost efficiency
+- **Scope Limits** - Guidance on max items per council to prevent token explosion
+
 ### 2.3.0 - 2026-01-14
 - Initial public release
 - Complete debate and quick workflows
diff --git a/Packs/pai-council-skill/src/skills/Council/Config.md b/Packs/pai-council-skill/src/skills/Council/Config.md
new file mode 100644
index 000000000..91fc32af9
--- /dev/null
+++ b/Packs/pai-council-skill/src/skills/Council/Config.md
@@ -0,0 +1,126 @@
+# Council Configuration
+
+Default configuration for Council debates. These values can be overridden per-session by specifying config inline with the council request.
+
+## Debate Settings
+
+| Setting | Default | Description |
+|---------|---------|-------------|
+| `rounds` | 2-3 (adaptive) | Round 3 runs only if convergence is low after Round 2 |
+| `agents` | 4 | Default: Architect, Designer, Engineer, Researcher |
+| `output_mode` | `deliberative` | Options: `deliberative` (conversational), `patchlist` (structured changes) |
+| `file_output` | `true` | Write each round to `~/.claude/MEMORY/` |
+
+## Model Tiering
+
+| Phase | Model | Rationale |
+|-------|-------|-----------|
+| Round 1 | sonnet | Initial positions need nuance |
+| Round 2 | haiku | Responses are shorter, more constrained |
+| Round 3 | sonnet | Synthesis needs quality |
+| Final Synthesis | sonnet | Orchestrator synthesis |
+
+To use a specific model, add `model: "haiku"` or `model: "sonnet"` to the Task call.
+
+## Adaptive Round 3
+
+Round 3 is **conditional**. After Round 2, the orchestrator evaluates:
+
+**Run Round 3 if ANY of these are true:**
+1. Agents express explicit disagreement in Round 2
+2. Round 2 responses identify BLOCKING issues
+3. The topic involves security, compliance, or irreversible decisions
+4. User explicitly requested 3 rounds
+
+**Skip Round 3 if ALL of these are true:**
+1. High convergence detected (majority of agents agree on core recommendation)
+2. No BLOCKING items identified
+3. No unresolved contradictions
+4. Topic is exploratory, not normative
+
+When skipping Round 3, output:
+```markdown
+**Round 3 skipped:** High convergence detected. Proceeding to synthesis.
+```
+
+## Output Modes
+
+### Deliberative (default)
+Conversational format for architectural debates, design decisions, exploratory discussions.
+
+```markdown
+### Round N: [Phase Name]
+
+**🏛️ Architect (Serena):**
+[Prose response, 50-150 words]
+```
+
+### Patchlist
+Structured format for TaskSpec reviews, document improvements, code changes.
+
+```markdown
+### Round N: [Phase Name]
+
+**🏛️ Architect (Serena):**
+
+**BLOCKING:**
+- B1: [issue] → [proposed change] | Affects: [specs]
+
+**HIGH:**
+- H1: [issue] → [proposed change]
+
+**MEDIUM/LOW:**
+- M1: [issue]
+```
+
+To invoke patchlist mode: `"Council (patchlist): Review these TaskSpecs..."`
+
+## File Output Location
+
+When `file_output` is enabled (default), rounds are written to:
+
+```
+~/.claude/MEMORY/STATE/council-sessions/{session-id}/
+├── metadata.json       # Topic, members, timing, config
+├── round-1.md          # All Round 1 responses
+├── round-2.md          # All Round 2 responses
+├── round-3.md          # Round 3 (if run)
+└── convergence.json    # Convergence assessment after Round 2
+```
+
+After synthesis, the final output is archived to:
+```
+~/.claude/MEMORY/RESEARCH/YYYY-MM/
+└── YYYY-MM-DD-HHMMSS_COUNCIL_{topic-slug}.md
+```
+
+This enables:
+- **Recovery** from interrupted sessions (partial round data preserved)
+- **Cross-session context** (prior council findings available)
+- **Resistance to context compaction** (durable external state)
+
+## Scope Limits
+
+To prevent token explosion and rate-limit issues:
+
+| Limit | Value | Rationale |
+|-------|-------|-----------|
+| Max items per council | 5 | TaskSpecs, documents, or topics |
+| Max agents | 7 | Diminishing returns beyond this |
+| Max words per response | 150 | Prevents verbose agents |
+
+For larger reviews, run multiple councils with a shared context preamble.
+
+## Recovery Mode
+
+If a council session is interrupted (rate limit, timeout, crash), use the Recovery workflow:
+
+```
+"Council recovery: Resume session {session-id}"
+```
+
+See `Workflows/Recovery.md` for details.
+
+---
+
+**Last Updated:** 2026-02-02
diff --git a/Packs/pai-council-skill/src/skills/Council/SKILL.md b/Packs/pai-council-skill/src/skills/Council/SKILL.md
index 06d018bb2..1fbbb93c3 100755
--- a/Packs/pai-council-skill/src/skills/Council/SKILL.md
+++ b/Packs/pai-council-skill/src/skills/Council/SKILL.md
@@ -12,13 +12,20 @@ science_cycle_time: meso
 
 If this directory exists, load and apply any PREFERENCES.md, configurations, or resources found there. These override default behavior. If the directory does not exist, proceed with skill defaults.
 
+## Configuration
+
+Load `Config.md` for default settings:
+- **Adaptive rounds:** 2-3 based on convergence
+- **File output:** Enabled by default (writes to `~/.claude/MEMORY/`)
+- **Model tiering:** sonnet for Round 1/3, haiku for Round 2
+- **Output modes:** `deliberative` (default) or `patchlist`
+
 # Council Skill
 
 Multi-agent debate system where specialized agents discuss topics in rounds, respond to each other's points, and surface insights through intellectual friction.
 
 **Key Differentiator from RedTeam:** Council is collaborative-adversarial (debate to find best path), while RedTeam is purely adversarial (attack the idea). Council produces visible conversation transcripts; RedTeam produces steelman + counter-argument.
 
-
 ## Voice Notification
 
 **When executing a workflow, do BOTH:**
@@ -50,42 +57,53 @@ Running the **WorkflowName** workflow from the **Council** skill...
 
 | Trigger | Workflow |
 |---------|----------|
-| Full structured debate (3 rounds, visible transcript) | `Workflows/Debate.md` |
+| Full structured debate (2-3 rounds, visible transcript) | `Workflows/Debate.md` |
 | Quick consensus check (1 round, fast) | `Workflows/Quick.md` |
+| Resume interrupted session | `Workflows/Recovery.md` |
 | Pure adversarial analysis | RedTeam skill |
 
 ## Quick Reference
 
 | Workflow | Purpose | Rounds | Output |
 |----------|---------|--------|--------|
-| **DEBATE** | Full structured discussion | 3 | Complete transcript + synthesis |
+| **DEBATE** | Full structured discussion | 2-3 (adaptive) | Complete transcript + synthesis |
 | **QUICK** | Fast perspective check | 1 | Initial positions only |
+| **RECOVERY** | Resume interrupted session | Remaining | Continue from last checkpoint |
 
 ## Context Files
 
 | File | Content |
 |------|---------|
+| `Config.md` | Default settings, model tiering, output modes |
 | `CouncilMembers.md` | Agent roles, perspectives, voice mapping |
-| `RoundStructure.md` | Three-round debate structure and timing |
+| `RoundStructure.md` | Adaptive 2-3 round structure and timing |
 | `OutputFormat.md` | Transcript format templates |
 
 ## Core Philosophy
 
 **Origin:** Best decisions emerge from diverse perspectives challenging each other. Not just collecting opinions - genuine intellectual friction where experts respond to each other's actual points.
 
-**Speed:** Parallel execution within rounds, sequential between rounds. A 3-round debate of 4 agents = 12 agent calls but only 3 sequential waits. Complete in 30-90 seconds.
+**Speed:** Parallel execution within rounds, sequential between rounds. A 2-round debate of 4 agents = 8 agent calls but only 2 sequential waits. Complete in 20-60 seconds.
+
+**Resilience:** File-first output means session state survives context compaction and rate limits.
 
 ## Examples
 
 ```
 "Council: Should we use WebSockets or SSE?"
--> Invokes DEBATE workflow -> 3-round transcript
+-> Invokes DEBATE workflow -> 2-3 round transcript
 
 "Quick council check: Is this API design reasonable?"
 -> Invokes QUICK workflow -> Fast perspectives
 
 "Council with security: Evaluate this auth approach"
 -> DEBATE with Security agent added
+
+"Council (patchlist): Review these specifications"
+-> DEBATE with structured output format
+
+"Council recovery: Resume session 20260202-143052-a1b2c3d4"
+-> RECOVERY workflow -> Continue from checkpoint
 ```
 
 ## Integration
@@ -101,7 +119,9 @@ Running the **WorkflowName** workflow from the **Council** skill...
 2. Add domain-specific experts as needed (security for auth, etc.)
 3. Review the transcript - insights are in the responses, not just positions
 4. Trust multi-agent convergence when it occurs
+5. Use patchlist mode for specification reviews
+6. For large reviews (5+ items), run multiple councils with shared context
 
 ---
 
-**Last Updated:** 2025-12-20
+**Last Updated:** 2026-02-02
diff --git a/Packs/pai-council-skill/src/skills/Council/Workflows/Debate.md b/Packs/pai-council-skill/src/skills/Council/Workflows/Debate.md
index e6db22fc4..a88b31394 100755
--- a/Packs/pai-council-skill/src/skills/Council/Workflows/Debate.md
+++ b/Packs/pai-council-skill/src/skills/Council/Workflows/Debate.md
@@ -1,14 +1,43 @@
 # Debate Workflow
 
-Full structured multi-agent debate with 3 rounds and visible transcript.
+Full structured multi-agent debate with 2-3 adaptive rounds, file-first output, and visible transcript.
 
 ## Prerequisites
 
 - Topic or question to debate
 - Optional: Custom council members (default: architect, designer, engineer, researcher)
+- Optional: Output mode (`deliberative` or `patchlist`)
+
+## Configuration
+
+Load settings from `Config.md`. Key defaults:
+- **Rounds:** 2-3 (adaptive based on convergence)
+- **File output:** Enabled (writes to `~/.claude/MEMORY/`)
+- **Model tiering:** Round 1 sonnet, Round 2 haiku, Round 3/Synthesis sonnet
 
 ## Execution
 
+### Step 0: Initialize Session
+
+Generate a session ID and create session directory:
+
+```bash
+SESSION_ID=$(date +%Y%m%d-%H%M%S)-$(openssl rand -hex 4)
+mkdir -p ~/.claude/MEMORY/STATE/council-sessions/$SESSION_ID
+```
+
+Write `metadata.json`:
+```json
+{
+  "session_id": "{SESSION_ID}",
+  "topic": "{topic}",
+  "started_at": "{ISO timestamp}",
+  "members": ["Architect", "Designer", "Engineer", "Researcher"],
+  "output_mode": "deliberative",
+  "rounds_planned": "adaptive"
+}
+```
+
 ### Step 1: Announce the Council
 
 Output the debate header:
@@ -16,13 +45,15 @@ Output the debate header:
 ```markdown
 ## Council Debate: [Topic]
 
+**Session ID:** {SESSION_ID}
 **Council Members:** [List agents participating]
-**Rounds:** 3 (Positions → Responses → Synthesis)
+**Rounds:** 2-3 (adaptive based on convergence)
+**Output:** ~/.claude/MEMORY/STATE/council-sessions/{SESSION_ID}/
 ```
 
 ### Step 2: Round 1 - Initial Positions
 
-Launch 4 parallel Task calls (one per council member).
+Launch 4 parallel Task calls (one per council member). Use `model: "sonnet"`.
 
 **Each agent prompt includes:**
 ```
@@ -47,6 +78,12 @@ Your perspective focuses on: [agent's domain]
 - **engineer**: Implementation reality, tech debt, maintenance burden, practical constraints
 - **researcher** (ClaudeResearcher): Data, precedent, external examples, what others have done
 
+**Write Round 1 to file:**
+```bash
+# Write combined Round 1 output to session directory
+echo "[Round 1 content]" > ~/.claude/MEMORY/STATE/council-sessions/$SESSION_ID/round-1.md
+```
+
 **Output each response as it completes:**
 ```markdown
 ### Round 1: Initial Positions
@@ -66,7 +103,7 @@ Your perspective focuses on: [agent's domain]
 
 ### Step 3: Round 2 - Responses & Challenges
 
-Launch 4 parallel Task calls with Round 1 transcript included.
+Launch 4 parallel Task calls with Round 1 transcript included. Use `model: "haiku"`.
 
 **Each agent prompt includes:**
 ```
@@ -89,6 +126,11 @@ Now respond to the other council members:
 The value is in genuine intellectual friction—engage with their actual arguments.
 ```
 
+**Write Round 2 to file:**
+```bash
+echo "[Round 2 content]" > ~/.claude/MEMORY/STATE/council-sessions/$SESSION_ID/round-2.md
+```
+
 **Output:**
 ```markdown
 ### Round 2: Responses & Challenges
@@ -106,9 +148,36 @@ The value is in genuine intellectual friction—engage with their actual argumen
 [Response referencing others' points]
 ```
 
-### Step 4: Round 3 - Synthesis
+### Step 3.5: Convergence Check (Adaptive Round 3)
+
+After Round 2, evaluate whether Round 3 is needed.
+
+**Run Round 3 if ANY of these are true:**
+- Agents express explicit disagreement ("I disagree with...", "I challenge...")
+- BLOCKING issues identified (security, compliance, irreversible decisions)
+- Topic requires normative decision (not just exploratory)
+- User explicitly requested 3 rounds
+
+**Skip Round 3 if ALL of these are true:**
+- High convergence (majority of agents agree on core recommendation)
+- No BLOCKING items identified
+- No unresolved contradictions
+- Topic is exploratory
+
+Write convergence assessment:
+```bash
+# Example: high convergence, skipping Round 3
+echo '{"converged": true, "reason": "4/4 agents agree on core recommendation", "proceed_to_round3": false}' > ~/.claude/MEMORY/STATE/council-sessions/$SESSION_ID/convergence.json
+```
+
+If skipping Round 3:
+```markdown
+**Round 3 skipped:** High convergence detected. Proceeding to synthesis.
+```
+
+### Step 4: Round 3 - Synthesis (Conditional)
 
-Launch 4 parallel Task calls with Round 1 + Round 2 transcripts.
+If Round 3 is needed, launch 4 parallel Task calls with Round 1 + Round 2 transcripts. Use `model: "sonnet"`.
 
 **Each agent prompt includes:**
 ```
@@ -130,6 +199,11 @@ Final synthesis from your perspective:
 Be honest about remaining disagreements—forced consensus is worse than acknowledged tension.
 ```
 
+**Write Round 3 to file:**
+```bash
+echo "[Round 3 content]" > ~/.claude/MEMORY/STATE/council-sessions/$SESSION_ID/round-3.md
+```
+
 **Output:**
 ```markdown
 ### Round 3: Synthesis
@@ -155,7 +229,7 @@ After all rounds complete, synthesize the debate:
 ### Council Synthesis
 
 **Areas of Convergence:**
-- [Points where 3+ agents agreed]
+- [Points where majority of agents agreed]
 - [Shared concerns or recommendations]
 
 **Remaining Disagreements:**
@@ -166,6 +240,31 @@ After all rounds complete, synthesize the debate:
 [Based on convergence and weight of arguments, the recommended approach is...]
 ```
 
+### Step 6: Archive to RESEARCH
+
+After synthesis, archive the complete session:
+
+```bash
+TOPIC_SLUG=$(echo "[topic]" | tr '[:upper:]' '[:lower:]' | tr ' ' '-' | cut -c1-50)
+TIMESTAMP=$(date +%Y-%m-%d-%H%M%S)
+MONTH=$(date +%Y-%m)
+mkdir -p ~/.claude/MEMORY/RESEARCH/$MONTH
+# Explicit file ordering (round-3.md may not exist if skipped)
+cat ~/.claude/MEMORY/STATE/council-sessions/$SESSION_ID/round-1.md \
+    ~/.claude/MEMORY/STATE/council-sessions/$SESSION_ID/round-2.md \
+    ~/.claude/MEMORY/STATE/council-sessions/$SESSION_ID/round-3.md 2>/dev/null \
+    > ~/.claude/MEMORY/RESEARCH/$MONTH/${TIMESTAMP}_COUNCIL_${TOPIC_SLUG}.md
+```
+
+Update metadata.json with completion status (merge with existing):
+```json
+{
+  "completed_at": "2026-02-02T23:55:39Z",
+  "rounds_completed": 2,
+  "archived_to": "~/.claude/MEMORY/RESEARCH/2026-02/2026-02-02-235539_COUNCIL_example-topic.md"
+}
+```
+
 ## Custom Council Members
 
 If user specifies custom members, adjust accordingly:
@@ -177,25 +276,58 @@ If user specifies custom members, adjust accordingly:
 
 ## Agent Type Mapping
 
-| Council Role | Task subagent_type | Personality Reference |
-|--------------|-------------------|----------------------|
-| Architect | Architect | Serena Blackwood |
-| Designer | Designer | Aditi Sharma |
-| Engineer | Engineer | Marcus Webb |
-| Researcher | PerplexityResearcher | Ava Chen |
-| Security | Pentester | Rook Blackburn |
-| Intern | Intern | Dev Patel |
-| Writer | (use Intern with writer prompt) | Emma Hartley |
+| Council Role | Task subagent_type | Model | Personality Reference |
+|--------------|-------------------|-------|----------------------|
+| Architect | Architect | sonnet | Serena Blackwood |
+| Designer | Designer | sonnet | Aditi Sharma |
+| Engineer | Engineer | sonnet | Marcus Webb |
+| Researcher | PerplexityResearcher | sonnet | Ava Chen |
+| Security | Pentester | sonnet | Rook Blackburn |
+| Intern | Intern | haiku | Dev Patel |
+| Writer | (use Intern with writer prompt) | haiku | Emma Hartley |
 
 ## Timing
 
 - Round 1: ~10-20 seconds (parallel)
 - Round 2: ~10-20 seconds (parallel)
-- Round 3: ~10-20 seconds (parallel)
+- Convergence check: ~1 second
+- Round 3 (if needed): ~10-20 seconds (parallel)
 - Synthesis: ~5 seconds
 
-**Total: 30-90 seconds for full debate**
+**Total: 20-60 seconds** (2 rounds) or **30-90 seconds** (3 rounds)
+
+## Output Modes
+
+### Deliberative (default)
+Standard conversational format. Use for architectural debates, design decisions, exploratory discussions.
+
+### Patchlist
+Structured format for specification reviews. Invoke with: `"Council (patchlist): Review..."`
+
+In patchlist mode, instruct agents to structure responses as:
+
+```markdown
+**BLOCKING:**
+- B1: [issue] → [proposed change]
+
+**HIGH:**
+- H1: [issue] → [proposed change]
+
+**MEDIUM/LOW:**
+- M1: [issue]
+```
+
+## Interruption Handling
+
+If the council is interrupted (rate limit, timeout):
+1. Session state is preserved in `~/.claude/MEMORY/STATE/council-sessions/{SESSION_ID}/`
+2. Use Recovery workflow to resume: `"Council recovery: Resume session 20260202-235539-a1b2c3d4"` (replace with your actual session ID)
+3. See `Workflows/Recovery.md` for details
 
 ## Done
 
-Debate complete. The transcript shows the full intellectual journey from initial positions through challenges to synthesis.
+Debate complete. The transcript is archived to `~/.claude/MEMORY/RESEARCH/` and shows the full intellectual journey from initial positions through challenges to synthesis.
+
+---
+
+**Last Updated:** 2026-02-02
diff --git a/Packs/pai-council-skill/src/skills/Council/Workflows/Recovery.md b/Packs/pai-council-skill/src/skills/Council/Workflows/Recovery.md
new file mode 100644
index 000000000..5e0b102c7
--- /dev/null
+++ b/Packs/pai-council-skill/src/skills/Council/Workflows/Recovery.md
@@ -0,0 +1,189 @@
+# Recovery Workflow
+
+Resume an interrupted Council session or rerun with partial prior results.
+
+## When to Use
+
+- Session was interrupted by rate limit, timeout, or context compaction
+- Partial round outputs were saved before interruption
+- You want to continue without re-running completed work
+
+## Prerequisites
+
+- Session ID or session directory path
+- Prior round outputs (partial or complete)
+- Knowledge of which agents completed / which didn't
+
+## Recovery Modes
+
+### Full Rerun
+
+Re-run all agents for all remaining rounds. Ignores prior partial outputs.
+
+**Use when:**
+- Only 1-2 agents completed before interruption
+- Significant time has passed (context may have drifted)
+- User explicitly prefers fresh run
+
+**Invoke:** `"Council recovery (full rerun): {topic}"`
+
+### Partial Rerun
+
+Keep completed agent outputs, only rerun missing agents.
+
+**Use when:**
+- 3+ agents completed Round 1
+- Interruption happened mid-round
+- Time-sensitive and want to preserve progress
+
+**Invoke:** `"Council recovery (partial): Resume session {session-id}"`
+
+## Execution
+
+### Step 1: Load Session State
+
+**Session ID validation:** Before using a session ID in file paths, validate it matches the expected format `YYYYMMDD-HHMMSS-[hex]` (e.g., `20260202-235539-a1b2c3d4`). Reject IDs containing path traversal characters (`/`, `\`, `..`).
+
+Read the session directory:
+```
+~/.claude/MEMORY/STATE/council-sessions/{session-id}/
+├── metadata.json     # Topic, members, config
+├── round-1.md        # Partial or complete
+├── round-2.md        # Partial or complete (if exists)
+└── convergence.json  # (if exists)
+```
+
+Parse `metadata.json` to determine:
+- Which round was interrupted
+- Which agents completed that round
+- Which agents need to be re-run
+
+### Step 2: Announce Recovery
+
+```markdown
+## Council Recovery: [Topic]
+
+**Session ID:** {session-id}
+**Mode:** [Full Rerun | Partial Rerun]
+**Prior Progress:**
+- Round 1: [Complete | Partial - {N}/4 agents]
+- Round 2: [Complete | Partial | Not started]
+- Round 3: [Complete | Partial | Not started]
+
+**Recovery Plan:**
+- [List of agents to rerun]
+- [Rounds to complete]
+```
+
+### Step 3a: Full Rerun
+
+If full rerun mode:
+
+1. Archive prior session state to `~/.claude/MEMORY/RESEARCH/` with `_INTERRUPTED` suffix
+2. Start fresh session with new session ID
+3. Follow standard Debate workflow from Round 1
+
+### Step 3b: Partial Rerun
+
+If partial rerun mode:
+
+1. Load completed agent outputs from prior round
+2. Launch only the missing agents with this context:
+
+```
+You are [Agent Name], [brief role description].
+
+COUNCIL RECOVERY - ROUND {N}: [PHASE NAME]
+
+Topic: [The topic being debated]
+
+Context: This is a recovery from an interrupted session. The following agents already completed Round {N}:
+
+[Prior agent outputs from this round]
+
+Your task:
+- Provide your Round {N} response
+- You may reference the prior agents' points
+- Stay consistent with the established discussion
+- [Standard round instructions]
+```
+
+3. Merge recovered outputs with new outputs
+4. Continue to next round
+
+### Step 4: Resume Normal Flow
+
+After recovery round completes:
+- Update session state files
+- Continue with next round (or synthesis if Round 3 complete)
+- Follow standard adaptive Round 3 logic
+
+### Step 5: Mark Recovery Complete
+
+After synthesis:
+```markdown
+**Recovery Complete:** Session resumed from Round {N}. Full transcript archived.
+```
+
+Archive final output to:
+```
+~/.claude/MEMORY/RESEARCH/YYYY-MM/YYYY-MM-DD-HHMMSS_COUNCIL_{topic-slug}_RECOVERED.md
+```
+
+## Diff-Only Mode (Partial Rerun Optimization)
+
+For partial reruns where 3+ agents completed, instruct recovered agents:
+
+```
+You are joining a council session in progress. The following perspectives have already been shared:
+
+[Prior agent outputs]
+
+Your task is to add NET-NEW insights only:
+- Do not repeat points already made
+- Focus on what the prior agents missed
+- State your unique perspective
+- If you agree with prior points, briefly acknowledge and move on
+```
+
+This reduces token usage while preserving the intellectual friction value.
+
+## Error Handling
+
+| Scenario | Action |
+|----------|--------|
+| Session directory not found | Ask user for session ID or start fresh |
+| metadata.json corrupted | Start fresh, note what was lost |
+| Round files missing | Treat as partial, rerun missing rounds |
+| All agents completed but synthesis missing | Skip to synthesis step |
+
+## Example Recovery Flow
+
+```
+User: "Council recovery: Resume session abc123"
+
+System reads: ~/.claude/MEMORY/STATE/council-sessions/abc123/
+- metadata.json shows: Round 1 interrupted, 1/4 agents complete
+- round-1.md contains: Aditi (Designer) completed
+
+System response:
+## Council Recovery: [Topic from metadata]
+
+**Session ID:** abc123
+**Mode:** Partial Rerun
+**Prior Progress:**
+- Round 1: Partial - 1/4 agents (Aditi complete)
+- Round 2: Not started
+
+**Recovery Plan:**
+- Rerun Round 1 for: Architect (Serena), Engineer (Marcus), Researcher (Ava)
+- Complete Round 2
+- Evaluate for Round 3
+- Synthesize
+
+Proceeding with recovery...
+```
+
+---
+
+**Last Updated:** 2026-02-02