agileflow 3.4.2 → 3.4.3

package/src/core/commands/code/logic.md

@@ -10,6 +10,8 @@ compact_context:
     - "CRITICAL: Confidence scoring: CONFIRMED (2+ agree), LIKELY (1 with evidence), INVESTIGATE (1 weak)"
     - "MUST parse arguments: TARGET (file/dir), DEPTH (quick/deep/ultradeep), FOCUS (edge/invariant/flow/type/race/all)"
     - "Pass consensus all analyzer outputs, let it synthesize the final report"
+    - "DEPTH GATE: ultradeep/extreme MUST spawn tmux sessions via spawn-audit-sessions.js — NEVER deploy in-process"
+    - "Use check-sessions.js to monitor spawned tmux sessions — NEVER write custom polling scripts"
   state_fields:
     - target_path
     - depth
@@ -71,7 +73,7 @@ Deploy multiple specialized logic analyzers in parallel to find bugs, then synth
 | Argument | Values | Default | Description |
 |----------|--------|---------|-------------|
 | TARGET | file/directory | `.` | What to analyze |
-| DEPTH | quick, deep, ultradeep | quick | quick = high-impact only, deep = comprehensive, ultradeep = separate tmux sessions |
+| DEPTH | quick, deep, ultradeep, extreme | quick | quick = high-impact only, deep = comprehensive, ultradeep = separate tmux sessions, extreme = partitioned tmux sessions |
 | FOCUS | edge,invariant,flow,type,race,all | all | Which analyzers to deploy |
 | MODEL | haiku, sonnet, opus | haiku | Model for analyzer subagents. Default preserves existing behavior. |
 
@@ -100,12 +102,12 @@ FOCUS = all (default) or comma-separated list
    Parse the JSON output to get `traceId`. Example: `{"ok":true,"traceId":"abc123ef",...}`
 4. Wait for all analyzers to complete:
    ```bash
-   node .agileflow/scripts/lib/tmux-audit-monitor.js wait TRACE_ID --timeout=1800
+   node .agileflow/scripts/check-sessions.js wait TRACE_ID --timeout=1800
    ```
    - Exit 0 = all complete (JSON results on stdout)
    - Exit 1 = timeout (partial results on stdout, `missing` array shows what's left)
-   - To check progress without blocking: `node .agileflow/scripts/lib/tmux-audit-monitor.js status TRACE_ID`
-   - To retry stalled analyzers: `node .agileflow/scripts/lib/tmux-audit-monitor.js retry TRACE_ID`
+   - To check progress without blocking: `node .agileflow/scripts/check-sessions.js status TRACE_ID`
+   - To retry stalled analyzers: `node .agileflow/scripts/check-sessions.js retry TRACE_ID`
 5. Parse `results` array from the JSON output. Pass all findings to consensus coordinator (same as deep mode).
 6. If tmux unavailable (spawn exits code 2), fall back to `DEPTH=deep` with warning
 
@@ -122,7 +124,7 @@ Partition-based multi-agent audit. Instead of 1 analyzer per tmux window, the co
    ```bash
    node .agileflow/scripts/spawn-audit-sessions.js --audit=logic --target=TARGET --depth=extreme --partitions=dir1,dir2,dir3 --model=MODEL --json
    ```
-4. Wait and collect results (same as ultradeep - use tmux-audit-monitor.js)
+4. Wait and collect results (same as ultradeep - use check-sessions.js)
 5. Run consensus on combined results from all partitions
 
 **PARTITIONS argument** (only used with DEPTH=extreme):
@@ -144,7 +146,21 @@ Partition-based multi-agent audit. Instead of 1 analyzer per tmux window, the co
 | `race` | logic-analyzer-race only |
 | `edge,type` | logic-analyzer-edge + logic-analyzer-type |
 
-### STEP 2: Deploy Analyzers in Parallel
+---
+
+### DEPTH ROUTING GATE
+
+| DEPTH | Route |
+|-------|-------|
+| `quick` or `deep` | Continue to STEP 2 below |
+| `ultradeep` | STOP. Follow ULTRADEEP instructions above. Do NOT proceed to STEP 2. |
+| `extreme` | STOP. Follow EXTREME instructions above. Do NOT proceed to STEP 2. |
+
+**CRITICAL**: STEP 2 is for `quick`/`deep` ONLY. For `ultradeep`/`extreme`, the analyzers run in separate tmux sessions — NOT in-process via Task calls. If you deploy Task calls for ultradeep/extreme, you are doing it wrong. Follow the spawn-audit-sessions.js workflow above, then skip to the consensus step with the collected results.
+
+---
+
+### STEP 2: Deploy Analyzers in Parallel (quick/deep ONLY)
 
 **CRITICAL**: Deploy ALL selected analyzers in a SINGLE message with multiple Task calls.
 

package/src/core/commands/code/performance.md

@@ -10,6 +10,8 @@ compact_context:
     - "CRITICAL: Confidence scoring: CONFIRMED (2+ agree), LIKELY (1 with evidence), INVESTIGATE (1 weak)"
     - "MUST parse arguments: TARGET (file/dir), DEPTH (quick/deep/ultradeep), FOCUS (queries|rendering|memory|bundle|compute|network|caching|assets|all)"
     - "Pass consensus all analyzer outputs, let it synthesize the final report"
+    - "DEPTH GATE: ultradeep/extreme MUST spawn tmux sessions via spawn-audit-sessions.js — NEVER deploy in-process"
+    - "Use check-sessions.js to monitor spawned tmux sessions — NEVER write custom polling scripts"
   state_fields:
     - target_path
     - depth
@@ -73,7 +75,7 @@ Deploy multiple specialized performance analyzers in parallel to find bottleneck
 | Argument | Values | Default | Description |
 |----------|--------|---------|-------------|
 | TARGET | file/directory | `.` | What to analyze |
-| DEPTH | quick, deep, ultradeep | quick | quick = core 5, deep = all 8, ultradeep = separate tmux sessions |
+| DEPTH | quick, deep, ultradeep, extreme | quick | quick = core 5, deep = all 8, ultradeep = separate tmux sessions, extreme = partitioned tmux sessions |
 | FOCUS | queries,rendering,memory,bundle,compute,network,caching,assets,all | all | Which analyzers to deploy |
 | MODEL | haiku, sonnet, opus | haiku | Model for analyzer subagents. Default preserves existing behavior. |
 
@@ -123,12 +125,12 @@ FOCUS = all (default) or comma-separated list
    Parse the JSON output to get `traceId`. Example: `{"ok":true,"traceId":"abc123ef",...}`
 4. Wait for all analyzers to complete:
    ```bash
-   node .agileflow/scripts/lib/tmux-audit-monitor.js wait TRACE_ID --timeout=1800
+   node .agileflow/scripts/check-sessions.js wait TRACE_ID --timeout=1800
    ```
    - Exit 0 = all complete (JSON results on stdout)
    - Exit 1 = timeout (partial results on stdout, `missing` array shows what's left)
-   - To check progress without blocking: `node .agileflow/scripts/lib/tmux-audit-monitor.js status TRACE_ID`
-   - To retry stalled analyzers: `node .agileflow/scripts/lib/tmux-audit-monitor.js retry TRACE_ID`
+   - To check progress without blocking: `node .agileflow/scripts/check-sessions.js status TRACE_ID`
+   - To retry stalled analyzers: `node .agileflow/scripts/check-sessions.js retry TRACE_ID`
 5. Parse `results` array from the JSON output. Pass all findings to consensus coordinator (same as deep mode).
 6. If tmux unavailable (spawn exits code 2), fall back to `DEPTH=deep` with warning
 
@@ -145,7 +147,7 @@ Partition-based multi-agent audit. Instead of 1 analyzer per tmux window, the co
    ```bash
    node .agileflow/scripts/spawn-audit-sessions.js --audit=performance --target=TARGET --depth=extreme --partitions=dir1,dir2,dir3 --model=MODEL --json
    ```
-4. Wait and collect results (same as ultradeep - use tmux-audit-monitor.js)
+4. Wait and collect results (same as ultradeep - use check-sessions.js)
 5. Run consensus on combined results from all partitions
 
 **PARTITIONS argument** (only used with DEPTH=extreme):
@@ -155,7 +157,21 @@ Partition-based multi-agent audit. Instead of 1 analyzer per tmux window, the co
 | `PARTITIONS=5` | AI creates exactly 5 partitions |
 | `PARTITIONS=src/auth,src/api,lib` | Use these exact directories |
 
-### STEP 2: Deploy Analyzers in Parallel
+---
+
+### DEPTH ROUTING GATE
+
+| DEPTH | Route |
+|-------|-------|
+| `quick` or `deep` | Continue to STEP 2 below |
+| `ultradeep` | STOP. Follow ULTRADEEP instructions above. Do NOT proceed to STEP 2. |
+| `extreme` | STOP. Follow EXTREME instructions above. Do NOT proceed to STEP 2. |
+
+**CRITICAL**: STEP 2 is for `quick`/`deep` ONLY. For `ultradeep`/`extreme`, the analyzers run in separate tmux sessions — NOT in-process via Task calls. If you deploy Task calls for ultradeep/extreme, you are doing it wrong. Follow the spawn-audit-sessions.js workflow above, then skip to the consensus step with the collected results.
+
+---
+
+### STEP 2: Deploy Analyzers in Parallel (quick/deep ONLY)
 
 **CRITICAL**: Deploy ALL selected analyzers in a SINGLE message with multiple Task calls.
 

package/src/core/commands/code/security.md

@@ -10,6 +10,8 @@ compact_context:
     - "CRITICAL: Confidence scoring: CONFIRMED (2+ agree), LIKELY (1 with evidence), INVESTIGATE (1 weak)"
     - "MUST parse arguments: TARGET (file/dir), DEPTH (quick/deep/ultradeep), FOCUS (injection|auth|authz|secrets|input|deps|infra|api|all)"
     - "Pass consensus all analyzer outputs, let it synthesize the final report"
+    - "DEPTH GATE: ultradeep/extreme MUST spawn tmux sessions via spawn-audit-sessions.js — NEVER deploy in-process"
+    - "Use check-sessions.js to monitor spawned tmux sessions — NEVER write custom polling scripts"
   state_fields:
     - target_path
     - depth
@@ -76,7 +78,7 @@ Deploy multiple specialized security vulnerability analyzers in parallel to find
 | Argument | Values | Default | Description |
 |----------|--------|---------|-------------|
 | TARGET | file/directory | `.` | What to analyze |
-| DEPTH | quick, deep, ultradeep | quick | quick = core 5, deep = all 8, ultradeep = separate tmux sessions |
+| DEPTH | quick, deep, ultradeep, extreme | quick | quick = core 5, deep = all 8, ultradeep = separate tmux sessions, extreme = partitioned tmux sessions |
 | FOCUS | injection,auth,authz,secrets,input,deps,infra,api,all | all | Which analyzers to deploy |
 | MODEL | haiku, sonnet, opus | haiku | Model for analyzer subagents. Default preserves existing behavior. |
 
@@ -126,12 +128,12 @@ FOCUS = all (default) or comma-separated list
    Parse the JSON output to get `traceId`. Example: `{"ok":true,"traceId":"abc123ef",...}`
 4. Wait for all analyzers to complete:
    ```bash
-   node .agileflow/scripts/lib/tmux-audit-monitor.js wait TRACE_ID --timeout=1800
+   node .agileflow/scripts/check-sessions.js wait TRACE_ID --timeout=1800
    ```
    - Exit 0 = all complete (JSON results on stdout)
    - Exit 1 = timeout (partial results on stdout, `missing` array shows what's left)
-   - To check progress without blocking: `node .agileflow/scripts/lib/tmux-audit-monitor.js status TRACE_ID`
-   - To retry stalled analyzers: `node .agileflow/scripts/lib/tmux-audit-monitor.js retry TRACE_ID`
+   - To check progress without blocking: `node .agileflow/scripts/check-sessions.js status TRACE_ID`
+   - To retry stalled analyzers: `node .agileflow/scripts/check-sessions.js retry TRACE_ID`
 5. Parse `results` array from the JSON output. Pass all findings to consensus coordinator (same as deep mode).
 6. If tmux unavailable (spawn exits code 2), fall back to `DEPTH=deep` with warning
 
@@ -148,7 +150,7 @@ Partition-based multi-agent audit. Instead of 1 analyzer per tmux window, the co
    ```bash
    node .agileflow/scripts/spawn-audit-sessions.js --audit=security --target=TARGET --depth=extreme --partitions=dir1,dir2,dir3 --model=MODEL --json
    ```
-4. Wait and collect results (same as ultradeep - use tmux-audit-monitor.js)
+4. Wait and collect results (same as ultradeep - use check-sessions.js)
 5. Run consensus on combined results from all partitions
 
 **PARTITIONS argument** (only used with DEPTH=extreme):
@@ -158,7 +160,21 @@ Partition-based multi-agent audit. Instead of 1 analyzer per tmux window, the co
 | `PARTITIONS=5` | AI creates exactly 5 partitions |
 | `PARTITIONS=src/auth,src/api,lib` | Use these exact directories |
 
-### STEP 2: Deploy Analyzers in Parallel
+---
+
+### DEPTH ROUTING GATE
+
+| DEPTH | Route |
+|-------|-------|
+| `quick` or `deep` | Continue to STEP 2 below |
+| `ultradeep` | STOP. Follow ULTRADEEP instructions above. Do NOT proceed to STEP 2. |
+| `extreme` | STOP. Follow EXTREME instructions above. Do NOT proceed to STEP 2. |
+
+**CRITICAL**: STEP 2 is for `quick`/`deep` ONLY. For `ultradeep`/`extreme`, the analyzers run in separate tmux sessions — NOT in-process via Task calls. If you deploy Task calls for ultradeep/extreme, you are doing it wrong. Follow the spawn-audit-sessions.js workflow above, then skip to the consensus step with the collected results.
+
+---
+
+### STEP 2: Deploy Analyzers in Parallel (quick/deep ONLY)
 
 **CRITICAL**: Deploy ALL selected analyzers in a SINGLE message with multiple Task calls.
 

package/src/core/commands/code/test.md

@@ -10,6 +10,8 @@ compact_context:
     - "CRITICAL: Confidence scoring: CONFIRMED (2+ agree), LIKELY (1 with evidence), INVESTIGATE (1 weak)"
     - "MUST parse arguments: TARGET (file/dir), DEPTH (quick/deep/ultradeep), FOCUS (coverage|fragility|mocking|assertions|structure|integration|maintenance|patterns|all)"
     - "Pass consensus all analyzer outputs, let it synthesize the final report"
+    - "DEPTH GATE: ultradeep/extreme MUST spawn tmux sessions via spawn-audit-sessions.js — NEVER deploy in-process"
+    - "Use check-sessions.js to monitor spawned tmux sessions — NEVER write custom polling scripts"
   state_fields:
     - target_path
     - depth
@@ -73,7 +75,7 @@ Deploy multiple specialized test quality analyzers in parallel to find weaknesse
 | Argument | Values | Default | Description |
 |----------|--------|---------|-------------|
 | TARGET | file/directory | `.` | What to analyze |
-| DEPTH | quick, deep, ultradeep | quick | quick = core 5, deep = all 8, ultradeep = separate tmux sessions |
+| DEPTH | quick, deep, ultradeep, extreme | quick | quick = core 5, deep = all 8, ultradeep = separate tmux sessions, extreme = partitioned tmux sessions |
 | FOCUS | coverage,fragility,mocking,assertions,structure,integration,maintenance,patterns,all | all | Which analyzers to deploy |
 | MODEL | haiku, sonnet, opus | haiku | Model for analyzer subagents. Default preserves existing behavior. |
 
@@ -123,12 +125,12 @@ FOCUS = all (default) or comma-separated list
    Parse the JSON output to get `traceId`. Example: `{"ok":true,"traceId":"abc123ef",...}`
 4. Wait for all analyzers to complete:
    ```bash
-   node .agileflow/scripts/lib/tmux-audit-monitor.js wait TRACE_ID --timeout=1800
+   node .agileflow/scripts/check-sessions.js wait TRACE_ID --timeout=1800
    ```
    - Exit 0 = all complete (JSON results on stdout)
    - Exit 1 = timeout (partial results on stdout, `missing` array shows what's left)
-   - To check progress without blocking: `node .agileflow/scripts/lib/tmux-audit-monitor.js status TRACE_ID`
-   - To retry stalled analyzers: `node .agileflow/scripts/lib/tmux-audit-monitor.js retry TRACE_ID`
+   - To check progress without blocking: `node .agileflow/scripts/check-sessions.js status TRACE_ID`
+   - To retry stalled analyzers: `node .agileflow/scripts/check-sessions.js retry TRACE_ID`
 5. Parse `results` array from the JSON output. Pass all findings to consensus coordinator (same as deep mode).
 6. If tmux unavailable (spawn exits code 2), fall back to `DEPTH=deep` with warning
 
@@ -145,7 +147,7 @@ Partition-based multi-agent audit. Instead of 1 analyzer per tmux window, the co
    ```bash
    node .agileflow/scripts/spawn-audit-sessions.js --audit=test --target=TARGET --depth=extreme --partitions=dir1,dir2,dir3 --model=MODEL --json
    ```
-4. Wait and collect results (same as ultradeep - use tmux-audit-monitor.js)
+4. Wait and collect results (same as ultradeep - use check-sessions.js)
 5. Run consensus on combined results from all partitions
 
 **PARTITIONS argument** (only used with DEPTH=extreme):
@@ -155,7 +157,21 @@ Partition-based multi-agent audit. Instead of 1 analyzer per tmux window, the co
 | `PARTITIONS=5` | AI creates exactly 5 partitions |
 | `PARTITIONS=src/auth,src/api,lib` | Use these exact directories |
 
-### STEP 2: Deploy Analyzers in Parallel
+---
+
+### DEPTH ROUTING GATE
+
+| DEPTH | Route |
+|-------|-------|
+| `quick` or `deep` | Continue to STEP 2 below |
+| `ultradeep` | STOP. Follow ULTRADEEP instructions above. Do NOT proceed to STEP 2. |
+| `extreme` | STOP. Follow EXTREME instructions above. Do NOT proceed to STEP 2. |
+
+**CRITICAL**: STEP 2 is for `quick`/`deep` ONLY. For `ultradeep`/`extreme`, the analyzers run in separate tmux sessions — NOT in-process via Task calls. If you deploy Task calls for ultradeep/extreme, you are doing it wrong. Follow the spawn-audit-sessions.js workflow above, then skip to the consensus step with the collected results.
+
+---
+
+### STEP 2: Deploy Analyzers in Parallel (quick/deep ONLY)
 
 **CRITICAL**: Deploy ALL selected analyzers in a SINGLE message with multiple Task calls.
 

package/src/core/commands/ideate/features.md

@@ -11,6 +11,7 @@ compact_context:
     - "CRITICAL: Confidence scoring: CONFIRMED (2+ agree), LIKELY (1 with evidence), SPECULATIVE (1 weak)"
     - "MUST parse arguments: TARGET (file/dir), DEPTH (quick/deep/ultradeep), FOCUS (features|ux|market|growth|integration|all), MODEL (haiku/sonnet/opus)"
     - "Pass consensus all analyzer outputs, let it synthesize the final report"
+    - "Use check-sessions.js to monitor spawned tmux sessions — NEVER write custom polling scripts"
   state_fields:
     - target_path
     - depth
@@ -143,12 +144,12 @@ FOCUS = all (default) or comma-separated list
    Parse the JSON output to get `traceId`. Example: `{"ok":true,"traceId":"abc123ef",...}`
 4. Wait for all analyzers to complete:
    ```bash
-   node .agileflow/scripts/lib/tmux-audit-monitor.js wait TRACE_ID --timeout=1800
+   node .agileflow/scripts/check-sessions.js wait TRACE_ID --timeout=1800
    ```
    - Exit 0 = all complete (JSON results on stdout)
    - Exit 1 = timeout (partial results on stdout, `missing` array shows what's left)
-   - To check progress without blocking: `node .agileflow/scripts/lib/tmux-audit-monitor.js status TRACE_ID`
-   - To retry stalled analyzers: `node .agileflow/scripts/lib/tmux-audit-monitor.js retry TRACE_ID`
+   - To check progress without blocking: `node .agileflow/scripts/check-sessions.js status TRACE_ID`
+   - To retry stalled analyzers: `node .agileflow/scripts/check-sessions.js retry TRACE_ID`
 5. Parse `results` array from the JSON output. Pass all findings to consensus coordinator (same as deep mode).
 6. If tmux unavailable (spawn exits code 2), fall back to `DEPTH=deep` with warning
 
@@ -165,7 +166,7 @@ Partition-based multi-agent audit. Instead of 1 analyzer per tmux window, the co
    ```bash
    node .agileflow/scripts/spawn-audit-sessions.js --audit=brainstorm --target=TARGET --depth=extreme --partitions=dir1,dir2,dir3 --model=MODEL --json
    ```
-4. Wait and collect results (same as ultradeep - use tmux-audit-monitor.js)
+4. Wait and collect results (same as ultradeep - use check-sessions.js)
 5. Run consensus on combined results from all partitions
 
 **PARTITIONS argument** (only used with DEPTH=extreme):

package/src/core/commands/ideate/new.md

@@ -15,6 +15,7 @@ compact_context:
     - "MUST estimate effort for each idea: High/Medium/Low impact"
     - "MUST assign IDEA-XXXX identifiers to all ideas and update ideation index"
     - "Optional: Generate stories for HIGH-confidence items (if OUTPUT=stories or both)"
+    - "Use check-sessions.js to monitor spawned tmux sessions — NEVER write custom polling scripts"
   state_fields:
     - scope
     - depth
@@ -364,12 +365,12 @@ Parse the input arguments:
    Parse the JSON output to get `traceId`. Example: `{"ok":true,"traceId":"abc123ef",...}`
 4. Wait for all analyzers to complete:
    ```bash
-   node .agileflow/scripts/lib/tmux-audit-monitor.js wait TRACE_ID --timeout=1800
+   node .agileflow/scripts/check-sessions.js wait TRACE_ID --timeout=1800
    ```
    - Exit 0 = all complete (JSON results on stdout)
    - Exit 1 = timeout (partial results on stdout, `missing` array shows what's left)
-   - To check progress without blocking: `node .agileflow/scripts/lib/tmux-audit-monitor.js status TRACE_ID`
-   - To retry stalled analyzers: `node .agileflow/scripts/lib/tmux-audit-monitor.js retry TRACE_ID`
+   - To check progress without blocking: `node .agileflow/scripts/check-sessions.js status TRACE_ID`
+   - To retry stalled analyzers: `node .agileflow/scripts/check-sessions.js retry TRACE_ID`
 5. Parse `results` array from the JSON output and proceed to STEP 4 (Synthesis).
 6. If tmux unavailable (spawn exits code 2), fall back to `DEPTH=deep` with warning
 
@@ -386,7 +387,7 @@ Partition-based multi-agent audit. Instead of 1 expert per tmux window, the code
    ```bash
    node .agileflow/scripts/spawn-audit-sessions.js --audit=ideate --target=TARGET --depth=extreme --partitions=dir1,dir2,dir3 --focus=FOCUS_KEYS --model=MODEL --json
    ```
-4. Wait and collect results (same as ultradeep - use tmux-audit-monitor.js)
+4. Wait and collect results (same as ultradeep - use check-sessions.js)
 5. Run synthesis on combined results from all partitions (proceed to STEP 4)
 
 **PARTITIONS argument** (only used with DEPTH=extreme):
@@ -521,11 +522,11 @@ Each expert gets its own Claude Code tmux session with full context window. Use
 
 ```bash
 # Check completion status
-node .agileflow/scripts/lib/tmux-audit-monitor.js status TRACE_ID
+node .agileflow/scripts/check-sessions.js status TRACE_ID
 # Wait for all to complete (blocks until done or timeout)
-node .agileflow/scripts/lib/tmux-audit-monitor.js wait TRACE_ID --timeout=1800
+node .agileflow/scripts/check-sessions.js wait TRACE_ID --timeout=1800
 # Collect whatever results are ready
-node .agileflow/scripts/lib/tmux-audit-monitor.js collect TRACE_ID
+node .agileflow/scripts/check-sessions.js collect TRACE_ID
 ```
 
 **For SCOPE-filtered ultradeep** (e.g., SCOPE=security):

package/src/core/commands/seo/audit.md

@@ -11,6 +11,8 @@ compact_context:
     - "MUST parse arguments: URL (required), DEPTH (quick/deep/ultradeep/extreme), MAX_PAGES (default 50)"
     - "Fetch homepage FIRST to detect business type before deploying analyzers"
     - "Pass all analyzer outputs to seo-consensus for final report"
+    - "DEPTH GATE: ultradeep/extreme MUST spawn tmux sessions via spawn-audit-sessions.js — NEVER deploy in-process"
+    - "Use check-sessions.js to monitor spawned tmux sessions — NEVER write custom polling scripts"
   state_fields:
     - target_url
     - depth
@@ -112,12 +114,12 @@ MAX_PAGES = 50 (default)
    Parse the JSON output to get `traceId`. Example: `{"ok":true,"traceId":"abc123ef",...}`
 4. Wait for all analyzers to complete:
    ```bash
-   node .agileflow/scripts/lib/tmux-audit-monitor.js wait TRACE_ID --timeout=1800
+   node .agileflow/scripts/check-sessions.js wait TRACE_ID --timeout=1800
    ```
    - Exit 0 = all complete (JSON results on stdout)
    - Exit 1 = timeout (partial results on stdout, `missing` array shows what's left)
-   - To check progress without blocking: `node .agileflow/scripts/lib/tmux-audit-monitor.js status TRACE_ID`
-   - To retry stalled analyzers: `node .agileflow/scripts/lib/tmux-audit-monitor.js retry TRACE_ID`
+   - To check progress without blocking: `node .agileflow/scripts/check-sessions.js status TRACE_ID`
+   - To retry stalled analyzers: `node .agileflow/scripts/check-sessions.js retry TRACE_ID`
 5. Parse `results` array from the JSON output. Pass all findings to consensus coordinator (same as deep mode).
 6. If tmux unavailable (spawn exits code 2), fall back to `DEPTH=deep` with warning
 
@@ -134,7 +136,7 @@ Partition-based multi-agent audit. Instead of auditing the entire site as one un
    ```bash
    node .agileflow/scripts/spawn-audit-sessions.js --audit=seo --target=URL --depth=extreme --partitions=/blog,/docs,/products --model=MODEL --json
    ```
-4. Wait and collect results (same as ultradeep - use tmux-audit-monitor.js)
+4. Wait and collect results (same as ultradeep - use check-sessions.js)
 5. Run consensus on combined results from all partitions
 
 **PARTITIONS argument** (only used with DEPTH=extreme):
@@ -144,6 +146,20 @@ Partition-based multi-agent audit. Instead of auditing the entire site as one un
 | `PARTITIONS=5` | AI creates exactly 5 page groups |
 | `PARTITIONS=/blog,/docs,/products` | Use these exact URL path prefixes |
 
+---
+
+### DEPTH ROUTING GATE
+
+| DEPTH | Route |
+|-------|-------|
+| `quick` or `deep` | Continue to STEP 2 below |
+| `ultradeep` | STOP. Follow ULTRADEEP instructions above. Do NOT proceed to STEP 2. |
+| `extreme` | STOP. Follow EXTREME instructions above. Do NOT proceed to STEP 2. |
+
+**CRITICAL**: STEPs 2-4 are for `quick`/`deep` ONLY. For `ultradeep`/`extreme`, the analyzers run in separate tmux sessions — NOT in-process via Task calls. If you deploy Task calls for ultradeep/extreme, you are doing it wrong. Follow the spawn-audit-sessions.js workflow above, then skip to the consensus step with the collected results.
+
+---
+
 If URL is missing, ask:
 ```xml
 <invoke name="AskUserQuestion">
@@ -174,7 +190,7 @@ Also fetch:
 - `{URL}/robots.txt` - crawl rules, sitemap location
 - `{URL}/sitemap.xml` - page inventory
 
-### STEP 3: Deploy 6 Analyzers in Parallel
+### STEP 3: Deploy 6 Analyzers in Parallel (quick/deep ONLY)
 
 **CRITICAL**: Deploy ALL 6 analyzers in a SINGLE message with multiple Task calls.
 

package/lib/claude-cli-bridge.js

@@ -1,215 +0,0 @@
-/**
- * Claude CLI Bridge
- *
- * Spawns Claude Code CLI as a subprocess and bridges communication
- * between the dashboard and Claude.
- */
-
-const { spawn } = require('child_process');
-const readline = require('readline');
-const path = require('path');
-
-/**
- * Creates a Claude CLI bridge for a session
- */
-function createClaudeBridge(options = {}) {
-  const {
-    cwd = process.cwd(),
-    onText,
-    onToolStart,
-    onToolResult,
-    onError,
-    onComplete,
-    onInit,
-  } = options;
-
-  let claudeProcess = null;
-  let isRunning = false;
-
-  /**
-   * Send a message to Claude and stream the response
-   */
-  async function sendMessage(content) {
-    if (isRunning) {
-      throw new Error('Claude is already processing a message');
-    }
-
-    isRunning = true;
-
-    return new Promise((resolve, reject) => {
-      // Spawn claude with streaming JSON output
-      claudeProcess = spawn(
-        'claude',
-        ['--print', '--output-format', 'stream-json', '--permission-mode', 'default', content],
-        {
-          cwd,
-          env: { ...process.env },
-          stdio: ['ignore', 'pipe', 'pipe'],
-        }
-      );
-
-      // Parse NDJSON output line by line
-      const rl = readline.createInterface({
-        input: claudeProcess.stdout,
-        crlfDelay: Infinity,
-      });
-
-      let fullResponse = '';
-      let toolUseBuffer = new Map();
-      let hadAssistantText = false;
-
-      rl.on('line', line => {
-        try {
-          const event = JSON.parse(line);
-          const result = processEvent(event, {
-            onText,
-            onToolStart,
-            onToolResult,
-            onInit,
-            toolUseBuffer,
-            hadAssistantText,
-          });
-
-          // Track if we've sent assistant text
-          if (result?.sentText) {
-            hadAssistantText = true;
-          }
-
-          // Accumulate text response
-          if (event.type === 'assistant' && event.message?.content) {
-            for (const block of event.message.content) {
-              if (block.type === 'text') {
-                fullResponse += block.text;
-              }
-            }
-          }
-          // Also accumulate from result (for slash commands)
-          if (event.type === 'result' && event.result) {
-            fullResponse = event.result;
-          }
-        } catch (err) {
-          // Skip non-JSON lines
-        }
-      });
-
-      claudeProcess.stderr.on('data', data => {
-        const errorText = data.toString();
-        if (onError) onError(errorText);
-      });
-
-      claudeProcess.on('close', code => {
-        isRunning = false;
-        claudeProcess = null;
-
-        if (code === 0) {
-          if (onComplete) onComplete(fullResponse);
-          resolve(fullResponse);
-        } else {
-          const error = new Error(`Claude exited with code ${code}`);
-          if (onError) onError(error.message);
-          reject(error);
-        }
-      });
-
-      claudeProcess.on('error', err => {
-        isRunning = false;
-        claudeProcess = null;
-        if (onError) onError(err.message);
-        reject(err);
-      });
-    });
-  }
-
-  /**
-   * Cancel the current Claude operation
-   */
-  function cancel() {
-    if (claudeProcess) {
-      claudeProcess.kill('SIGINT');
-      isRunning = false;
-      claudeProcess = null;
-    }
-  }
-
-  /**
-   * Check if Claude is currently processing
-   */
-  function isBusy() {
-    return isRunning;
-  }
-
-  return {
-    sendMessage,
-    cancel,
-    isBusy,
-  };
-}
-
-/**
- * Process a stream-json event from Claude CLI
- * @returns {{ sentText: boolean }} - Whether text was sent
- */
-function processEvent(event, handlers) {
-  const { onText, onToolStart, onToolResult, onInit, toolUseBuffer, hadAssistantText } = handlers;
-  let sentText = false;
-
-  switch (event.type) {
-    case 'system':
-      if (event.subtype === 'init' && onInit) {
-        onInit({
-          sessionId: event.session_id,
-          model: event.model,
-          tools: event.tools,
-        });
-      }
-      break;
-
-    case 'assistant':
-      if (event.message?.content) {
-        for (const block of event.message.content) {
-          if (block.type === 'text' && onText) {
-            // Stream text
-            onText(block.text, false);
-            sentText = true;
-          } else if (block.type === 'tool_use' && onToolStart) {
-            // Tool call started
-            toolUseBuffer.set(block.id, block);
-            onToolStart(block.id, block.name, block.input);
-          }
-        }
-      }
-      break;
-
-    case 'user':
-      // This is a tool result coming back
-      if (event.message?.content) {
-        for (const block of event.message.content) {
-          if (block.type === 'tool_result' && onToolResult) {
-            const toolUse = toolUseBuffer.get(block.tool_use_id);
-            onToolResult(block.tool_use_id, block.content, block.is_error, toolUse?.name);
-          }
-        }
-      }
-      break;
-
-    case 'result':
-      if (event.subtype === 'success') {
-        // For slash commands (no assistant text), send the result text
-        if (event.result && onText && !hadAssistantText) {
-          onText(event.result, false);
-          sentText = true;
-        }
-        // Signal completion
-        if (onText) {
-          onText('', true);
-        }
-      }
-      break;
-  }
-
-  return { sentText };
-}
-
-module.exports = {
-  createClaudeBridge,
-};