prizmkit 1.1.92 → 1.1.94

package/bundled/VERSION.json

@@ -1,5 +1,5 @@
 {
-  "frameworkVersion": "1.1.92",
-  "bundledAt": "2026-07-02T15:14:07.667Z",
-  "bundledFrom": "edf0ddb"
+  "frameworkVersion": "1.1.94",
+  "bundledAt": "2026-07-02T17:32:35.248Z",
+  "bundledFrom": "3cd68b5"
 }

package/bundled/dev-pipeline/.env.example

@@ -23,7 +23,6 @@
 
 # ─── Pipeline Execution ──────────────────────────────────────────────
 # MAX_RETRIES=3              # Max retry attempts per task before marking failed
-# SESSION_TIMEOUT=0          # Session timeout in seconds (0 = no limit)
 # VERBOSE=1                  # Verbose logging (1=on, 0=off)
 
 # ─── Feature Pipeline Only ────────────────────────────────────────────
@@ -45,6 +44,3 @@
 # CODEX_WAIT_STALE_KILL_THRESHOLD=3600  # Longer no-log window while Codex waits on subagents
 # CODEX_SUBAGENT_TIMEOUT_SECONDS=3300   # Codex subagent max runtime; defaults to wait threshold - 300
 # CB_SUBAGENT_STALE_KILL_THRESHOLD=3600 # Longer no-log window while CodeBuddy sub-agents are running
-# LOG_CLEANUP_ENABLED=1      # Periodic log cleanup (1=on, 0=off)
-# LOG_RETENTION_DAYS=14      # Delete logs older than N days
-# LOG_MAX_TOTAL_MB=1024      # Keep total logs under N MB via oldest-first cleanup

package/bundled/dev-pipeline/README.md

@@ -53,7 +53,6 @@ Main entry point. Drives the full feature development pipeline.
 | Option | Description |
 |--------|-------------|
 | `--dry-run` | Generate bootstrap prompt only, don't spawn AI session |
-| `--timeout N` | Override `SESSION_TIMEOUT` for this run (seconds) |
 | `--resume-phase N` | Resume from specific phase number |
 | `--no-reset` | Don't reset feature artifacts before running |
 
@@ -71,9 +70,6 @@ Main entry point. Drives the full feature development pipeline.
 
 # Resume from Phase 6 (implementation)
 ./.prizmkit/dev-pipeline/run-feature.sh run .prizmkit/plans/feature-list.json --resume-phase 6
-
-# With timeout per session
-SESSION_TIMEOUT=7200 ./.prizmkit/dev-pipeline/run-feature.sh run .prizmkit/plans/feature-list.json
 ```
 
 If `.prizmkit/plans/feature-list.json` path is omitted, defaults to `.prizmkit/plans/feature-list.json` in the project root.
@@ -140,7 +136,7 @@ Manages `run-feature.sh` as a background daemon process with PID tracking and lo
 ```bash
 ./.prizmkit/dev-pipeline/launch-feature-daemon.sh start .prizmkit/plans/feature-list.json \
   --mode standard \
-  --env "MAX_RETRIES=5 SESSION_TIMEOUT=3600 MODEL=claude-sonnet-4.6"
+  --env "MAX_RETRIES=5 MODEL=claude-sonnet-4.6"
 ```
 
 **Output (JSON on stdout for programmatic consumption):**
@@ -177,7 +173,7 @@ Equivalent to `run-feature.sh` but for the bug-fix pipeline.
 ./run-bugfix.sh help
 ```
 
-**Single-bug options:** `--dry-run`, `--timeout N` (same as `run-feature.sh`)
+**Single-bug options:** `--dry-run` (same as `run-feature.sh`)
 
 Processes bugs by: **severity** (critical > high > medium > low) then **priority** (high > medium > low).
 
@@ -342,7 +338,7 @@ python3 scripts/update-feature-status.py \
 | `complete` | `--feature-id` | Shortcut for manually marking completed |
 
 **Session status values (for `--session-status`):**
-`success`, `partial_resumable`, `partial_not_resumable`, `failed`, `crashed`, `timed_out`, `commit_missing`, `docs_missing`, `merge_conflict`
+`success`, `partial_resumable`, `partial_not_resumable`, `failed`, `crashed`, `commit_missing`, `docs_missing`, `merge_conflict`
 
 **Output (JSON to stdout):**
 
@@ -429,26 +425,6 @@ python3 scripts/detect-stuck.py \
 
 ---
 
-### `scripts/cleanup-logs.py` — Log File Cleanup
-
-Manages log file size by age and total size thresholds.
-
-```bash
-python3 scripts/cleanup-logs.py \
-  --state-dir <path> \
-  [--retention-days <n>] \
-  [--max-total-mb <n>] \
-  [--dry-run]
-```
-
-| Option | Default | Description |
-|--------|---------|-------------|
-| `--retention-days` | 14 | Delete files older than N days |
-| `--max-total-mb` | 1024 | If total > N MB, remove oldest first |
-| `--dry-run` | false | Show what would be deleted without deleting |
-
----
-
 ### `scripts/parse-stream-progress.py` — Real-Time Progress Parser
 
 Reads AI CLI `stream-json` output and extracts progress metrics.
@@ -533,7 +509,6 @@ Also exports: `log_info`, `log_warn`, `log_error`, `log_success` (with timestamp
 | Variable | Default | Used By | Description |
 |----------|---------|---------|-------------|
 | `MAX_RETRIES` | `3` | run-feature.sh | Max retry attempts per feature before marking as failed |
-| `SESSION_TIMEOUT` | `0` (none) | run-feature.sh, run-bugfix.sh | Timeout in seconds per AI CLI session. 0 = no timeout |
 | `PIPELINE_MODE` | (auto) | run-feature.sh, launch-feature-daemon.sh | Override mode for all features: `lite\|standard\|full` |
 | `ENABLE_CRITIC` | `false` | run-feature.sh, launch-feature-daemon.sh | Enable adversarial critic review: `true\|false` |
 | `DEV_BRANCH` | auto-generated | run-feature.sh | Custom git branch name (default: `dev/{feature-id}-{timestamp}`) |
@@ -555,21 +530,18 @@ Also exports: `log_info`, `log_warn`, `log_error`, `log_success` (with timestamp
 |----------|---------|---------|-------------|
 | `HEARTBEAT_INTERVAL` | `30` | run-feature.sh | Seconds between heartbeat log output |
 | `HEARTBEAT_STALE_THRESHOLD` | `600` | run-feature.sh | Seconds before a session is considered stale/stuck |
-| `LOG_CLEANUP_ENABLED` | `1` | run-feature.sh | Run log cleanup before pipeline execution |
-| `LOG_RETENTION_DAYS` | `14` | run-feature.sh | Delete logs older than N days |
-| `LOG_MAX_TOTAL_MB` | `1024` | run-feature.sh | Keep total log size under N MB |
 
 **Examples:**
 
 ```bash
 # All env vars in one invocation
-MAX_RETRIES=5 SESSION_TIMEOUT=7200 MODEL=claude-sonnet-4.6 VERBOSE=1 \
+MAX_RETRIES=5 MODEL=claude-sonnet-4.6 VERBOSE=1 \
   ./.prizmkit/dev-pipeline/run-feature.sh run .prizmkit/plans/feature-list.json
 
 # Via daemon with --env
 ./.prizmkit/dev-pipeline/launch-feature-daemon.sh start .prizmkit/plans/feature-list.json \
   --mode standard \
-  --env "MAX_RETRIES=5 SESSION_TIMEOUT=3600 AUTO_PUSH=1"
+  --env "MAX_RETRIES=5 AUTO_PUSH=1"
 ```
 
 ---
@@ -721,11 +693,10 @@ Features are executed in dependency order via DAG:
 
 1. **Bootstrap prompt** generated from feature spec, context, and acceptance criteria
 2. AI CLI spawned as background process with optional `--model` flag
-3. **Timeout watchdog** runs in parallel if `SESSION_TIMEOUT > 0`
-4. **Heartbeat monitor** prints progress every `HEARTBEAT_INTERVAL` seconds
-5. **Progress parser** (if stream-json supported) extracts phase/tool metrics
-6. On completion, `session-status.json` determines success/failure
-7. Feature status updated; on failure, retry count increments
+3. **Heartbeat monitor** prints progress every `HEARTBEAT_INTERVAL` seconds
+4. **Progress parser** (if stream-json supported) extracts phase/tool metrics
+5. On completion, `session-status.json` determines success/failure
+6. Feature status updated; on failure, retry count increments
 
 ### Heartbeat Output
 
@@ -882,7 +853,6 @@ run-bugfix.sh main loop
 |   +-- update-feature-status.py     # Feature state machine (8 actions)
 |   +-- update-bug-status.py         # Bug state machine (6 actions)
 |   +-- detect-stuck.py              # Detect stuck/stale features
-|   +-- cleanup-logs.py              # Age/size-based log cleanup
 |   +-- parse-stream-progress.py     # Real-time stream-json progress parser
 |
 +-- templates/

package/bundled/dev-pipeline/SCHEMA_ANALYSIS.md

@@ -344,7 +344,6 @@ pending, in_progress, completed, failed, skipped
 | `AI_CLI` | string | auto-detect | claude or cbc |
 | `MODEL` | string | (unset) | AI model override (per-task overrides this) |
 | `MAX_RETRIES` | integer | (not specified) | Retry attempts per task |
-| `SESSION_TIMEOUT` | integer | 0 | 0 = no limit |
 | `VERBOSE` | integer | (not specified) | 1=on, 0=off |
 | `ENABLE_CRITIC` | boolean | false | Adversarial review enable |
 | `PIPELINE_MODE` | string | auto-detect | lite/standard/full override |
@@ -356,9 +355,6 @@ pending, in_progress, completed, failed, skipped
 | `STALE_KILL_THRESHOLD` | integer | 900 | Auto-kill after N seconds without parent log progress |
 | `CODEX_WAIT_STALE_KILL_THRESHOLD` | integer | 3600 | Longer no-log stale window while Codex waits on subagents |
 | `CODEX_SUBAGENT_TIMEOUT_SECONDS` | integer | 3300 | Codex subagent max runtime |
-| `LOG_CLEANUP_ENABLED` | integer | 1 | Periodic cleanup |
-| `LOG_RETENTION_DAYS` | integer | 14 | Delete logs older than N days |
-| `LOG_MAX_TOTAL_MB` | integer | 1024 | Max total logs (MB) |
 
 **Priority:** Command-line env vars > .env file > defaults
 

package/bundled/dev-pipeline/assets/prizm-dev-team-integration.md

@@ -51,7 +51,6 @@ Located at `.prizmkit/dev-pipeline/scripts/`:
 | `check-session-status.py` | Read and validate session-status.json output |
 | `detect-stuck.py` | Detect stuck/hung pipeline sessions via heartbeat |
 | `parse-stream-progress.py` | Parse AI CLI output stream for progress tracking |
-| `cleanup-logs.py` | Clean up old pipeline logs and state files |
 | `utils.py` | Shared utility functions for pipeline scripts |
 
 ## Artifact Mapping
@@ -125,7 +124,6 @@ The agent MUST write `session-status.json` before exiting:
 | `status: "partial"`, `can_resume: false` | Retry from scratch |
 | `status: "failed"` | Retry (up to MAX_RETRIES) |
 | No status file (crash) | Treat as failed, retry |
-| Timeout (exit 124) | Treat as timed_out, retry |
 
 ## Team Naming Convention
 

package/bundled/dev-pipeline/launch-bugfix-daemon.sh

@@ -482,7 +482,6 @@ Examples:
 
 Environment Variables (pass via --env):
   MAX_RETRIES           Max retries per bug (default: 3)
-  SESSION_TIMEOUT       Session timeout in seconds (default: 0 = no limit)
   VERBOSE               Set to 1 for verbose AI CLI output
   HEARTBEAT_INTERVAL    Heartbeat log interval in seconds (default: 30)
   DEV_BRANCH            Custom dev branch name (default: auto-generated)

package/bundled/dev-pipeline/launch-feature-daemon.sh

@@ -106,7 +106,7 @@ cmd_start() {
             --env)
                 shift
                 if [[ $# -eq 0 ]]; then
-                    log_error "--env requires a value (e.g. --env \"MAX_RETRIES=5 SESSION_TIMEOUT=3600\")"
+                    log_error "--env requires a value (e.g. --env \"MAX_RETRIES=5\")"
                     exit 1
                 fi
                 env_overrides="$1"
@@ -609,7 +609,7 @@ Examples:
   ./launch-feature-daemon.sh start --features F-001,F-003,F-007  # Run specific features
   ./launch-feature-daemon.sh start --mode full              # Full mode for complex features
   ./launch-feature-daemon.sh start --critic                  # Enable adversarial critic review
-  ./launch-feature-daemon.sh start --env "MAX_RETRIES=5 SESSION_TIMEOUT=7200"
+  ./launch-feature-daemon.sh start --env "MAX_RETRIES=5"
   ./launch-feature-daemon.sh start .prizmkit/plans/feature-list.json --mode full --critic --env "VERBOSE=1"
   ./launch-feature-daemon.sh status                       # Check if running (JSON on stdout)
   ./launch-feature-daemon.sh logs --follow                # Live log tailing
@@ -619,7 +619,6 @@ Examples:
 
 Environment Variables (pass via --env):
   MAX_RETRIES           Max retries per feature (default: 3)
-  SESSION_TIMEOUT       Session timeout in seconds (default: 0 = no limit)
   VERBOSE               Set to 1 for verbose AI CLI output
   HEARTBEAT_INTERVAL    Heartbeat log interval in seconds (default: 30)
   DEV_BRANCH            Custom dev branch name (default: auto-generated)

package/bundled/dev-pipeline/launch-refactor-daemon.sh

@@ -483,7 +483,6 @@ Examples:
 
 Environment Variables (pass via --env):
   MAX_RETRIES              Max retries per refactor (default: 3)
-  SESSION_TIMEOUT          Session timeout in seconds (default: 0 = no limit)
   VERBOSE                  Set to 1 for verbose AI CLI output
   STRICT_BEHAVIOR_CHECK    Force full test suite after each refactor (default: 1)
   HEARTBEAT_INTERVAL       Heartbeat log interval in seconds (default: 30)

package/bundled/dev-pipeline/lib/heartbeat.sh

@@ -44,7 +44,6 @@ start_heartbeat() {
         local prev_max_size=0
         local prev_child_activity_signature=""
         local prev_progress_signature=""
-        local prev_error_loop_signature=""
         local stale_seconds=0
         local stale_reason="stale_session"
         while kill -0 "$cli_pid" 2>/dev/null; do
@@ -197,64 +196,24 @@ PY
                 fi
             fi
 
-            # Check for error-loop: agent is actively producing output but results are
-            # all read-offset errors or wasted calls. This is a stuck agent that appears
-            # "active" by log growth but is accomplishing nothing. Only a newly advancing
-            # error signature counts; stale historical errors must not turn later normal
-            # output into a stale kill.
-            local error_loop_detected=false
-            local error_loop_signature=""
-            if [[ $effective_stale_kill_threshold -gt 0 && $growth -gt 0 && -f "$progress_json" ]]; then
-                local error_loop_flag
-                error_loop_flag=$(python3 - "$progress_json" <<'PY' 2>/dev/null || true
-import json, sys
-try:
-    with open(sys.argv[1], encoding="utf-8") as fh:
-        progress = json.load(fh)
-except Exception:
-    raise SystemExit(0)
-errors = progress.get("errors", [])
-if isinstance(errors, list) and len(errors) >= 5:
-    recent = errors[-5:]
-    if all(isinstance(e, dict) and e.get("type") in ("read_offset_overflow", "wasted_call") for e in recent):
-        print(json.dumps(recent, sort_keys=True, separators=(",", ":")))
-PY
-                )
-                if [[ -n "$error_loop_flag" ]]; then
-                    error_loop_signature="$error_loop_flag"
-                    if [[ -n "$prev_error_loop_signature" && "$error_loop_signature" != "$prev_error_loop_signature" ]]; then
-                        error_loop_detected=true
-                    fi
-                fi
-            fi
-            prev_error_loop_signature="$error_loop_signature"
-
             # Track progress staleness. Parent sessions can sit in a wait/polling
             # tool while child transcripts or structured progress counters keep
             # advancing, so both child activity and progress.json state changes
-            # count as progress. Error loops bypass normal growth-as-progress
-            # because the log is only growing with repeated failed reads or wasted
-            # calls.
+            # count as progress.
             #
             # When the main session log has been truncated (e.g. by subagent stdout
             # redirection severing the file descriptor), use structured progress and
             # child activity as the primary progress signals — the parent process is
             # still running and children may still be producing output.
-            if [[ "$error_loop_detected" == "true" ]]; then
-                stale_reason="error_loop"
-                stale_seconds=$effective_stale_kill_threshold
-            elif [[ "$log_truncated" == "true" ]]; then
-                stale_reason="stale_session"
+            if [[ "$log_truncated" == "true" ]]; then
                 if [[ $progress_advanced -gt 0 || $child_growth -gt 0 ]]; then
                     stale_seconds=0
                 else
                     stale_seconds=$((stale_seconds + heartbeat_interval))
                 fi
             elif [[ $growth -le 0 && $child_growth -eq 0 && $progress_advanced -eq 0 ]]; then
-                stale_reason="stale_session"
                 stale_seconds=$((stale_seconds + heartbeat_interval))
             else
-                stale_reason="stale_session"
                 stale_seconds=0
             fi
 
@@ -338,13 +297,8 @@ PY
             # stale window to surface stuck agents promptly.
             if [[ $effective_stale_kill_threshold -gt 0 && $stale_seconds -ge $effective_stale_kill_threshold ]]; then
                 local stale_mins=$((stale_seconds / 60))
-                if [[ "$stale_reason" == "error_loop" ]]; then
-                    echo -e "  ${RED}[HEARTBEAT]${NC} ${mins}m${secs}s | log: ${size_display} | ${RED}STALE-KILL: repeated read-offset/wasted-call error loop (threshold: ${effective_stale_kill_threshold}s)${NC}"
-                    echo -e "  ${RED}[HEARTBEAT]${NC} Killing AI CLI process $cli_pid (error loop)..."
-                else
-                    echo -e "  ${RED}[HEARTBEAT]${NC} ${mins}m${secs}s | log: ${size_display} | ${RED}STALE-KILL: no progress for ${stale_mins}m (threshold: ${effective_stale_kill_threshold}s)${NC}"
-                    echo -e "  ${RED}[HEARTBEAT]${NC} Killing AI CLI process $cli_pid (stale session)..."
-                fi
+                echo -e "  ${RED}[HEARTBEAT]${NC} ${mins}m${secs}s | log: ${size_display} | ${RED}STALE-KILL: no progress for ${stale_mins}m (threshold: ${effective_stale_kill_threshold}s)${NC}"
+                echo -e "  ${RED}[HEARTBEAT]${NC} Killing AI CLI process $cli_pid (stale session)..."
                 # Write the marker before killing. Some CLIs exit quickly, and the
                 # parent runner may stop this heartbeat process immediately after
                 # wait(1) returns.

package/bundled/dev-pipeline/run-bugfix.sh

@@ -16,7 +16,6 @@ set -euo pipefail
 # Environment Variables:
 #   MAX_RETRIES           Max code retries per bug (default: 3)
 #   MAX_INFRA_RETRIES     Max infrastructure/provider retries per bug (default: 3)
-#   SESSION_TIMEOUT       Session timeout in seconds (default: 0 = no limit)
 #   AI_CLI                AI CLI command name (auto-detected: cbc, claude, or codex)
 #   CODEBUDDY_CLI         Legacy alias for AI_CLI (deprecated, use AI_CLI instead)
 #   PRIZMKIT_PLATFORM     Platform hint for custom AI_CLI wrappers: codebuddy, claude, or codex
@@ -24,9 +23,6 @@ set -euo pipefail
 #   HEARTBEAT_INTERVAL    Heartbeat log interval in seconds (default: 30)
 #   STALE_KILL_THRESHOLD   Auto-kill session after N seconds of no progress (default: 900)
 #   HEARTBEAT_STALE_THRESHOLD  Heartbeat stale threshold in seconds (default: 600)
-#   LOG_CLEANUP_ENABLED   Run periodic log cleanup (default: 1)
-#   LOG_RETENTION_DAYS    Delete logs older than N days (default: 14)
-#   LOG_MAX_TOTAL_MB      Keep total logs under N MB via oldest-first cleanup (default: 1024)
 #   DEV_BRANCH            Custom dev branch name (default: auto-generated bugfix/pipeline-{run_id})
 #   AUTO_PUSH             Auto-push to remote after successful bug fix (default: 0). Set to 1 to enable.
 #   STOP_ON_FAILURE       Stop pipeline after a task exhausts all retries (default: 0). Set to 1 to stop.
@@ -44,13 +40,9 @@ SCRIPTS_DIR="$SCRIPT_DIR/scripts"
 # Configuration
 MAX_RETRIES=${MAX_RETRIES:-3}
 MAX_INFRA_RETRIES=${MAX_INFRA_RETRIES:-3}
-SESSION_TIMEOUT=${SESSION_TIMEOUT:-0}
 HEARTBEAT_STALE_THRESHOLD=${HEARTBEAT_STALE_THRESHOLD:-600}
 HEARTBEAT_INTERVAL=${HEARTBEAT_INTERVAL:-30}
 STALE_KILL_THRESHOLD=${STALE_KILL_THRESHOLD:-900}
-LOG_CLEANUP_ENABLED=${LOG_CLEANUP_ENABLED:-1}
-LOG_RETENTION_DAYS=${LOG_RETENTION_DAYS:-14}
-LOG_MAX_TOTAL_MB=${LOG_MAX_TOTAL_MB:-1024}
 VERBOSE=${VERBOSE:-0}
 MODEL=${MODEL:-""}
 PRIZMKIT_EFFORT=${PRIZMKIT_EFFORT:-""}
@@ -117,13 +109,6 @@ spawn_and_wait_session() {
     start_progress_parser "$session_log" "$progress_json" "$SCRIPTS_DIR"
     local parser_pid="${_PARSER_PID:-}"
 
-    # Timeout watchdog
-    local watcher_pid=""
-    if [[ $SESSION_TIMEOUT -gt 0 ]]; then
-        ( sleep "$SESSION_TIMEOUT" && kill -TERM "$cli_pid" 2>/dev/null ) &
-        watcher_pid=$!
-    fi
-
     # Heartbeat monitor (with stale-kill protection)
     start_heartbeat "$cli_pid" "$session_log" "$progress_json" "$HEARTBEAT_INTERVAL" "$STALE_KILL_THRESHOLD"
     local heartbeat_pid="${_HEARTBEAT_PID:-}"
@@ -136,16 +121,13 @@ spawn_and_wait_session() {
         exit_code=$?
     fi
 
-    # Cleanup
-    [[ -n "$watcher_pid" ]] && kill "$watcher_pid" 2>/dev/null || true
+    # Clean up heartbeat and parser
     stop_heartbeat "$heartbeat_pid"
     stop_progress_parser "$parser_pid"
-    [[ -n "$watcher_pid" ]] && wait "$watcher_pid" 2>/dev/null || true
 
     # Recover session log from backup if truncated by subagent stdout redirection
     prizm_recover_session_log "$session_log"
 
-    [[ $exit_code -eq 143 ]] && exit_code=124
 
     # Check for stale-kill marker (heartbeat killed the process due to no progress)
     local stale_kill_marker="$session_dir/logs/stale-kill.json"
@@ -162,11 +144,7 @@ except Exception:
     pass
 PY
 )
-        if [[ "$stale_kill_reason" == "error_loop" ]]; then
-            log_warn "Session was killed by heartbeat monitor due to repeated read-offset/wasted-call errors"
-        else
-            log_warn "Session was stale-killed by heartbeat monitor (no progress for too long)"
-        fi
+        log_warn "Session was stale-killed by heartbeat monitor (no progress for too long)"
     fi
 
     local was_infra_error=false
@@ -192,19 +170,12 @@ PY
     project_root="$PROJECT_ROOT"
     local default_branch="$base_branch"
 
-    if [[ $exit_code -eq 124 ]]; then
-        log_warn "Session timed out after ${SESSION_TIMEOUT}s"
-        session_status="timed_out"
-    elif [[ "$was_infra_error" == true ]]; then
+    if [[ "$was_infra_error" == true ]]; then
         log_warn "Session failed due to AI CLI/provider infrastructure error"
         log_warn "Infrastructure errors are retried without consuming code retry budget"
         session_status="infra_error"
     elif [[ "$was_stale_killed" == true ]]; then
-        if [[ "$stale_kill_reason" == "error_loop" ]]; then
-            log_warn "Session killed due to repeated read-offset/wasted-call error loop"
-        else
-            log_warn "Session stale-killed (no progress for ${STALE_KILL_THRESHOLD}s)"
-        fi
+        log_warn "Session stale-killed (no progress for ${STALE_KILL_THRESHOLD}s)"
         log_warn "Stale-killed sessions are treated as failed; dev branch is preserved for inspection"
         session_status="crashed"
     elif [[ $exit_code -ne 0 ]]; then
@@ -399,28 +370,6 @@ check_dependencies() {
     prizm_ensure_git_repo "$_project_root"
 }
 
-run_log_cleanup() {
-    if [[ "$LOG_CLEANUP_ENABLED" != "1" ]]; then
-        return 0
-    fi
-
-    local cleanup_result
-    cleanup_result=$(python3 "$SCRIPTS_DIR/cleanup-logs.py" \
-        --state-dir "$STATE_DIR" \
-        --retention-days "$LOG_RETENTION_DAYS" \
-        --max-total-mb "$LOG_MAX_TOTAL_MB" 2>/dev/null) || {
-        log_warn "Log cleanup failed (continuing)"
-        return 0
-    }
-
-    local deleted reclaimed_kb
-    deleted=$(echo "$cleanup_result" | python3 -c "import sys,json; print(json.load(sys.stdin).get('deleted_files', 0))" 2>/dev/null || echo "0")
-    reclaimed_kb=$(echo "$cleanup_result" | python3 -c "import sys,json; print(int(json.load(sys.stdin).get('reclaimed_bytes', 0)/1024))" 2>/dev/null || echo "0")
-
-    if [[ "$deleted" -gt 0 ]]; then
-        log_info "Log cleanup: deleted $deleted files, reclaimed ${reclaimed_kb}KB"
-    fi
-}
 
 # ============================================================
 # run-one: Run a single bug fix
@@ -440,7 +389,6 @@ run_one() {
             --dry-run) dry_run=true; shift ;;
             --clean) do_clean=true; shift ;;
             --no-reset) no_reset=true; shift ;;
-            --timeout) shift; SESSION_TIMEOUT="${1:-0}"; shift ;;
             --max-infra-retries)
                 shift
                 if [[ $# -eq 0 ]]; then
@@ -491,7 +439,6 @@ run_one() {
     fi
 
     check_dependencies
-    run_log_cleanup
 
     # Initialize state if needed
     if [[ ! -f "$STATE_DIR/pipeline.json" ]]; then
@@ -877,7 +824,6 @@ main() {
     fi
 
     check_dependencies
-    run_log_cleanup
 
     # Initialize pipeline state if needed
     if [[ ! -f "$STATE_DIR/pipeline.json" ]]; then
@@ -926,11 +872,6 @@ main() {
     log_info "Bug fix list: $bug_list"
     log_info "Max code retries per bug: $MAX_RETRIES"
     log_info "Max infrastructure retries per bug: $MAX_INFRA_RETRIES"
-    if [[ $SESSION_TIMEOUT -gt 0 ]]; then
-        log_info "Session timeout: ${SESSION_TIMEOUT}s"
-    else
-        log_info "Session timeout: none"
-    fi
     log_info "AI CLI: $CLI_CMD (platform: $PLATFORM)"
     if [[ -n "${MODEL:-}" ]]; then
         log_info "Default Model: $MODEL"
@@ -1077,17 +1018,14 @@ DEPLOY_PROMPT_EOF
         fi
         echo -e "${BOLD}────────────────────────────────────────────────────${NC}"
 
-        # Do not auto-commit an arbitrary dirty baseline on the original branch.
-        # Only pipeline bookkeeping is committed explicitly; user/source WIP,
-        # submodule dirt, specs, and root-level hidden tool temporary worktrees
-        # must never be swept into main by a broad git add.
+        # Auto-commit any uncommitted workspace changes before starting the pipeline.
+        # This ensures a clean baseline without blocking execution.
         local _dirty_files=""
         _dirty_files=$(prizm_git_status_non_bookkeeping "$_proj_root" "$STATE_DIR" "$bug_list" || true)
         if [[ -n "$_dirty_files" ]]; then
-            log_error "Dirty working tree detected before $bug_id; refusing to commit it on $_ORIGINAL_BRANCH."
-            log_error "Commit/stash your changes or start from a clean branch, then rerun the pipeline."
-            printf '%s\n' "$_dirty_files" | sed 's/^/  /'
-            break
+            log_info "Auto-committing workspace changes before $bug_id..."
+            prizm_git_add_all_safe "$_proj_root"
+            git -C "$_proj_root" commit --no-verify -m "chore: workspace snapshot ready for run $bug_id" 2>/dev/null || true
         fi
 
         local _bookkeeping_dirty=""
@@ -1280,7 +1218,6 @@ show_help() {
     echo "  --dry-run                   Generate bootstrap prompt only, don't spawn session"
     echo "  --clean                     Delete artifacts and reset before running"
     echo "  --no-reset                  Skip status reset (preserve retry count)"
-    echo "  --timeout N                 Session timeout in seconds (default: 0 = no limit)"
     echo "  --max-infra-retries N       Max infrastructure/provider retries (default: 3)"
     echo "  --mode <lite|standard|full> Override pipeline mode"
     echo "  --critic                    Enable adversarial critic review"
@@ -1292,7 +1229,6 @@ show_help() {
     echo "Environment Variables:"
     echo "  MAX_RETRIES           Max code retries per bug (default: 3)"
     echo "  MAX_INFRA_RETRIES     Max infrastructure/provider retries per bug (default: 3)"
-    echo "  SESSION_TIMEOUT       Session timeout in seconds (default: 0 = no limit)"
     echo "  MODEL                 Default AI model (overridden by per-bug model in bug list)"
     echo "  PRIZMKIT_EFFORT       AI reasoning effort (low|medium|high|xhigh|max; max is Claude Code only)"
     echo "  PIPELINE_MODE         Default pipeline mode: lite|standard|full (overridden by --mode)"
@@ -1302,9 +1238,6 @@ show_help() {
     echo "  HEARTBEAT_INTERVAL    Heartbeat log interval in seconds (default: 30)"
     echo "  STALE_KILL_THRESHOLD  Auto-kill session after N seconds of no progress (default: 900)"
     echo "  STOP_ON_FAILURE       Stop pipeline when a task exhausts retries: 0|1 (default: 0)"
-    echo "  LOG_CLEANUP_ENABLED   Run log cleanup before execution (default: 1)"
-    echo "  LOG_RETENTION_DAYS    Delete logs older than N days (default: 14)"
-    echo "  LOG_MAX_TOTAL_MB      Keep total logs under N MB (default: 1024)"
     echo ""
     echo "Examples:"
     echo "  ./run-bugfix.sh run                                    # Run all bugs"
@@ -1312,7 +1245,6 @@ show_help() {
     echo "  ./run-bugfix.sh run B-001 --dry-run                    # Inspect generated prompt"
     echo "  ./run-bugfix.sh run B-001 --clean                      # Clean artifacts + reset + run"
     echo "  ./run-bugfix.sh run B-001 --no-reset                   # Retry without resetting status"
-    echo "  ./run-bugfix.sh run B-001 --timeout 3600               # 1h timeout"
     echo "  ./run-bugfix.sh status                                 # Show status"
     echo "  MAX_RETRIES=5 ./run-bugfix.sh run                      # Custom retries"
 }