prizmkit 1.1.72 → 1.1.76

package/bundled/VERSION.json

@@ -1,5 +1,5 @@
 {
-  "frameworkVersion": "1.1.72",
-  "bundledAt": "2026-06-11T12:58:26.870Z",
-  "bundledFrom": "099177b"
+  "frameworkVersion": "1.1.76",
+  "bundledAt": "2026-06-13T04:58:49.240Z",
+  "bundledFrom": "17f8257"
 }

package/bundled/agents/prizm-dev-team-dev.md

@@ -1,6 +1,6 @@
 ---
 name: prizm-dev-team-dev
-description: PrizmKit-integrated module implementer (multi-instance). Follows /prizmkit-implement workflow with TDD, marks tasks [x] in plan.md Tasks section, works within assigned Git worktrees. Use when implementing specific feature modules.
+description: PrizmKit-integrated module implementer (multi-instance). Follows /prizmkit-implement workflow with TDD, marks tasks [x] in plan.md Tasks section. Works directly in the repository; uses git worktree only if explicitly instructed. Use when implementing specific feature modules.
 tools: Read, Write, Edit, Bash, Glob, Grep, TaskCreate, TaskGet, TaskUpdate, TaskList, SendMessage
 disallowedTools: Agent
 model: inherit
@@ -86,6 +86,16 @@ DEV-17: DO NOT re-read source files already listed in context-snapshot.md Sectio
 DEV-18: When tests fail, run `$TEST_CMD 2>&1 | tee /tmp/test-out.txt` ONCE, then grep `/tmp/test-out.txt` for failure details. Never re-run the full test suite just to apply a different grep filter to its output.
 DEV-19: Before writing any `.prizmkit/prizm-docs/` file, check if it exists. If it exists: only update durable fields (KEY_FILES, INTERFACES, DEPENDENCIES, file counts, RULES, TRAPS, DECISIONS) — never overwrite the full file. Never add CHANGELOG, UPDATED/date fields, or workflow metadata. Only create new L2 docs for sub-modules you are actively creating in this session.
 DEV-20: Internal tracking IDs are not product copy. Before writing UI text or UI-copy assertions, translate references like `F-003 guard` into product-language behavior such as `the high-risk guard`. Add regression coverage when a feature touches user-visible text.
+DEV-21: Before any Read with offset + limit, compute offset + limit. If the last tool_result for this file shows it has N lines, offset MUST be < N. Never request an offset >= known file length.
+DEV-22: When Edit fails with 'String to replace not found':
+        1. STOP editing immediately — do NOT retry the same Edit.
+        2. Run `grep -n` to locate the exact line of the target text.
+        3. Read with offset = max(grep_line - 20, 0), limit = 50.
+        4. Copy the exact text from the Read result into the Edit old_string.
+        5. Never guess or extrapolate an offset — grep first, then read, then edit.
+DEV-23: If 3 consecutive Read calls to the same file return 'shorter than provided offset' or 'Wasted call', STOP all work on that file. Send ESCALATION with the file path, current line count, and the offsets you attempted. The orchestrator will provide the correct content.
+DEV-24: Before editing a large file (>1000 lines), verify you know its current line count from the most recent tool_result. Old line counts from earlier turns may be stale if you have since edited the file.
+DEV-25: After every 3 successful Edit operations on a file, run the relevant test command for that file once to validate your changes compile and behave correctly. Do not defer all testing to the end.
 ```
 
 ### Workflow

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

@@ -44,6 +44,7 @@
 # STALE_KILL_THRESHOLD=900   # Auto-kill after N seconds without parent log progress (0 = disabled)
 # 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/lib/common.sh

@@ -338,6 +338,41 @@ prizm_detect_cli_and_platform() {
     export PRIZMKIT_PLATFORM="$PLATFORM"
 }
 
+# Validate PRIZMKIT_EFFORT against the detected platform's supported values.
+# Exits with a friendly error if the value is unsupported.
+prizm_validate_effort() {
+    [[ -n "${PRIZMKIT_EFFORT:-}" ]] || return 0
+
+    local session_platform
+    session_platform="$(_prizm_session_platform)"
+
+    local allowed_values
+    case "$session_platform" in
+        claude)    allowed_values="low, medium, high, xhigh, max" ;;
+        codex)     allowed_values="low, medium, high, xhigh" ;;
+        *)         allowed_values="low, medium, high, xhigh" ;;
+    esac
+
+    local valid=0
+    local val
+    for val in $allowed_values; do
+        val="${val%,}"
+        [[ "$val" == "$PRIZMKIT_EFFORT" ]] && valid=1 && break
+    done
+
+    if [[ "$valid" -ne 1 ]]; then
+        local platform_label="$session_platform"
+        [[ "$session_platform" == "claude" ]] && platform_label="Claude Code"
+        [[ "$session_platform" == "codebuddy" ]] && platform_label="CodeBuddy"
+        [[ "$session_platform" == "codex" ]] && platform_label="Codex"
+
+        log_error "PRIZMKIT_EFFORT='$PRIZMKIT_EFFORT' is not supported by the detected CLI ($platform_label)."
+        log_error "  Supported values for $platform_label: $allowed_values"
+        log_error "  Set PRIZMKIT_EFFORT to one of the above, or unset it to use the CLI default."
+        exit 1
+    fi
+}
+
 # Start an AI CLI session in the background.
 # Usage: prizm_start_ai_session <prompt_path> <log_path> <model>
 # Sets PRIZM_AI_PID to the spawned process PID. Do not call this function via
@@ -380,6 +415,9 @@ prizm_start_ai_session() {
             if [[ "$USE_STREAM_JSON" == "true" ]]; then
                 claude_args+=(--output-format stream-json)
             fi
+            if [[ -n "${PRIZMKIT_EFFORT:-}" ]]; then
+                claude_args+=(--effort "$PRIZMKIT_EFFORT")
+            fi
             if [[ -n "$model" ]]; then
                 claude_args+=(--model "$model")
             fi
@@ -399,6 +437,9 @@ prizm_start_ai_session() {
             if [[ -n "$model" ]]; then
                 codex_args+=(--model "$model")
             fi
+            if [[ -n "${PRIZMKIT_EFFORT:-}" ]]; then
+                codex_args+=(--config "model_reasoning_effort=$PRIZMKIT_EFFORT")
+            fi
             "$CLI_CMD" "${codex_args[@]}" - < "$prompt_path" > "$log_path" 2>&1 &
             ;;
         *)
@@ -409,6 +450,9 @@ prizm_start_ai_session() {
             if [[ "$USE_STREAM_JSON" == "true" ]]; then
                 cb_args+=(--output-format stream-json)
             fi
+            if [[ -n "${PRIZMKIT_EFFORT:-}" ]]; then
+                cb_args+=(--effort "$PRIZMKIT_EFFORT")
+            fi
             if [[ -n "$model" ]]; then
                 cb_args+=(--model "$model")
             fi
@@ -900,6 +944,9 @@ prizm_run_ai_session() {
     case "$session_platform" in
         claude)
             local claude_args=(-p "$(cat "$prompt_path")" --dangerously-skip-permissions)
+            if [[ -n "${PRIZMKIT_EFFORT:-}" ]]; then
+                claude_args+=(--effort "$PRIZMKIT_EFFORT")
+            fi
             if [[ -n "$model" ]]; then
                 claude_args+=(--model "$model")
             fi
@@ -919,10 +966,16 @@ prizm_run_ai_session() {
             if [[ -n "$model" ]]; then
                 codex_args+=(--model "$model")
             fi
+            if [[ -n "${PRIZMKIT_EFFORT:-}" ]]; then
+                codex_args+=(--config "model_reasoning_effort=$PRIZMKIT_EFFORT")
+            fi
             "$CLI_CMD" "${codex_args[@]}" - < "$prompt_path" > "$log_path" 2>&1
             ;;
         *)
             local cb_args=(--print -y)
+            if [[ -n "${PRIZMKIT_EFFORT:-}" ]]; then
+                cb_args+=(--effort "$PRIZMKIT_EFFORT")
+            fi
             if [[ -n "$model" ]]; then
                 cb_args+=(--model "$model")
             fi

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

@@ -90,9 +90,102 @@ PY
             fi
             prev_child_activity_signature="$child_activity_signature"
 
+            local effective_stale_kill_threshold="$stale_kill_threshold"
+            if [[ $stale_kill_threshold -gt 0 && -f "$progress_json" ]]; then
+                local extended_threshold
+                extended_threshold=$(python3 - "$progress_json" "$stale_kill_threshold" <<'PY' 2>/dev/null || true
+import json
+import os
+import sys
+
+progress_path = sys.argv[1]
+base_threshold = int(sys.argv[2])
+
+with open(progress_path, "r", encoding="utf-8") as fh:
+    progress = json.load(fh)
+
+spawn_count = 0
+for tool in progress.get("tool_calls", []):
+    if isinstance(tool, dict) and tool.get("name") in ("spawn_agent", "Agent", "TaskCreate"):
+        try:
+            spawn_count += int(tool.get("count", 0))
+        except (TypeError, ValueError):
+            pass
+
+# Also check the subagent_spawn_count field (set by _record_cb_agent_tool_call)
+if not spawn_count:
+    spawn_count = int(progress.get("subagent_spawn_count", 0))
+
+fmt = progress.get("event_format", "")
+
+# Codex: current_tool == "wait" means parent is blocked on spawn_agent completion
+if (
+    fmt == "codex-json"
+    and progress.get("current_tool") == "wait"
+    and spawn_count > 0
+):
+    configured = os.environ.get("CODEX_WAIT_STALE_KILL_THRESHOLD", "")
+    try:
+        wait_threshold = int(configured)
+    except ValueError:
+        wait_threshold = max(base_threshold * 4, 3600)
+    if wait_threshold > base_threshold:
+        print(wait_threshold)
+
+# CodeBuddy: Agent tool blocks synchronously; Task* tools imply bg agents.
+# Extend the stale window when sub-agents have been spawned so the heartbeat
+# doesn't kill the parent while children are still running.
+if (
+    fmt == "stream-json"
+    and spawn_count > 0
+    and progress.get("cb_session_id", "")
+):
+    configured = os.environ.get("CB_SUBAGENT_STALE_KILL_THRESHOLD", "")
+    try:
+        cb_threshold = int(configured)
+    except ValueError:
+        cb_threshold = max(base_threshold * 4, 3600)
+    if cb_threshold > base_threshold:
+        print(cb_threshold)
+PY
+                )
+                if [[ "$extended_threshold" =~ ^[0-9]+$ && "$extended_threshold" -gt "$stale_kill_threshold" ]]; then
+                    effective_stale_kill_threshold="$extended_threshold"
+                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.
+            local error_loop_detected=false
+            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("error_loop")
+PY
+                )
+                if [[ "$error_loop_flag" == "error_loop" ]]; then
+                    error_loop_detected=true
+                fi
+            fi
+
             # Track progress staleness. Parent sessions can sit in a wait/polling
             # tool while child transcripts keep growing, so child activity counts.
-            if [[ $growth -eq 0 && $child_growth -eq 0 ]]; then
+            # Error loops bypass normal growth-as-progress because the log is only
+            # growing with repeated failed reads or wasted calls.
+            if [[ "$error_loop_detected" == "true" ]]; then
+                stale_seconds=$effective_stale_kill_threshold
+            elif [[ $growth -eq 0 && $child_growth -eq 0 ]]; then
                 stale_seconds=$((stale_seconds + heartbeat_interval))
             else
                 stale_seconds=0
@@ -132,47 +225,6 @@ PY
                 status_icon="${YELLOW}⏸${NC}"
             fi
 
-            local effective_stale_kill_threshold="$stale_kill_threshold"
-            if [[ $stale_kill_threshold -gt 0 && -f "$progress_json" ]]; then
-                local codex_wait_threshold
-                codex_wait_threshold=$(python3 - "$progress_json" "$stale_kill_threshold" <<'PY' 2>/dev/null || true
-import json
-import os
-import sys
-
-progress_path = sys.argv[1]
-base_threshold = int(sys.argv[2])
-
-with open(progress_path, "r", encoding="utf-8") as fh:
-    progress = json.load(fh)
-
-spawn_count = 0
-for tool in progress.get("tool_calls", []):
-    if isinstance(tool, dict) and tool.get("name") == "spawn_agent":
-        try:
-            spawn_count += int(tool.get("count", 0))
-        except (TypeError, ValueError):
-            pass
-
-if (
-    progress.get("event_format") == "codex-json"
-    and progress.get("current_tool") == "wait"
-    and spawn_count > 0
-):
-    configured = os.environ.get("CODEX_WAIT_STALE_KILL_THRESHOLD", "")
-    try:
-        wait_threshold = int(configured)
-    except ValueError:
-        wait_threshold = max(base_threshold * 4, 3600)
-    if wait_threshold > base_threshold:
-        print(wait_threshold)
-PY
-                )
-                if [[ "$codex_wait_threshold" =~ ^[0-9]+$ && "$codex_wait_threshold" -gt "$stale_kill_threshold" ]]; then
-                    effective_stale_kill_threshold="$codex_wait_threshold"
-                fi
-            fi
-
             # Fatal provider/runtime errors are terminal; do not wait for the
             # stale window when progress.json already proves the model cannot
             # continue (for example context_too_large).

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

@@ -51,6 +51,7 @@ LOG_RETENTION_DAYS=${LOG_RETENTION_DAYS:-14}
 LOG_MAX_TOTAL_MB=${LOG_MAX_TOTAL_MB:-1024}
 VERBOSE=${VERBOSE:-0}
 MODEL=${MODEL:-""}
+PRIZMKIT_EFFORT=${PRIZMKIT_EFFORT:-""}
 DEV_BRANCH=${DEV_BRANCH:-""}
 AUTO_PUSH=${AUTO_PUSH:-0}
 STOP_ON_FAILURE=${STOP_ON_FAILURE:-0}
@@ -59,6 +60,9 @@ ENABLE_DEPLOY=${ENABLE_DEPLOY:-0}
 # Source shared common helpers (CLI/platform detection + logs + deps)
 prizm_detect_cli_and_platform
 
+# Validate PRIZMKIT_EFFORT early (fail-fast before any sessions are spawned)
+prizm_validate_effort
+
 # Source shared heartbeat library
 source "$SCRIPT_DIR/lib/heartbeat.sh"
 
@@ -1197,6 +1201,7 @@ show_help() {
     echo "  MAX_RETRIES           Max 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)"
     echo "  ENABLE_CRITIC         Enable/disable critic: true|false|1|0 (overridden by --critic/--no-critic)"
     echo "  AI_CLI                AI CLI command name (auto-detected: cbc, claude, or codex)"

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

@@ -54,6 +54,7 @@ LOG_RETENTION_DAYS=${LOG_RETENTION_DAYS:-14}
 LOG_MAX_TOTAL_MB=${LOG_MAX_TOTAL_MB:-1024}
 VERBOSE=${VERBOSE:-0}
 MODEL=${MODEL:-""}
+PRIZMKIT_EFFORT=${PRIZMKIT_EFFORT:-""}
 DEV_BRANCH=${DEV_BRANCH:-""}
 AUTO_PUSH=${AUTO_PUSH:-0}
 STOP_ON_FAILURE=${STOP_ON_FAILURE:-0}
@@ -62,6 +63,9 @@ ENABLE_DEPLOY=${ENABLE_DEPLOY:-0}
 # Source shared common helpers (CLI/platform detection + logs + deps)
 prizm_detect_cli_and_platform
 
+# Validate PRIZMKIT_EFFORT early (fail-fast before any sessions are spawned)
+prizm_validate_effort
+
 # Source shared heartbeat library
 source "$SCRIPT_DIR/lib/heartbeat.sh"
 
@@ -1483,6 +1487,7 @@ show_help() {
     echo "  SESSION_TIMEOUT       Session timeout in seconds (default: 0 = no limit)"
     echo "  AI_CLI                AI CLI command name (auto-detected: cbc, claude, or codex)"
     echo "  MODEL                 AI model ID (e.g. claude-opus-4.6, claude-sonnet-4.6, claude-haiku-4.5)"
+    echo "  PRIZMKIT_EFFORT       AI reasoning effort (low|medium|high|xhigh|max; max is Claude Code only)"
     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 "  HEARTBEAT_STALE_THRESHOLD  Heartbeat stale threshold in seconds (default: 600)"

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

@@ -67,11 +67,15 @@ HEARTBEAT_INTERVAL=${HEARTBEAT_INTERVAL:-30}
 STALE_KILL_THRESHOLD=${STALE_KILL_THRESHOLD:-900}
 VERBOSE=${VERBOSE:-0}
 MODEL=${MODEL:-""}
+PRIZMKIT_EFFORT=${PRIZMKIT_EFFORT:-""}
 AUTO_PUSH=${AUTO_PUSH:-0}
 
 # Source shared common helpers (CLI/platform detection + logs + deps)
 prizm_detect_cli_and_platform
 
+# Validate PRIZMKIT_EFFORT early (fail-fast before any sessions are spawned)
+prizm_validate_effort
+
 # Source shared heartbeat library
 source "$SCRIPT_DIR/lib/heartbeat.sh"
 
@@ -106,6 +110,7 @@ Options (for 'run' command):
 Environment Variables:
   SESSION_TIMEOUT       Session timeout in seconds (default: 0 = no limit)
   MODEL                 AI model to use
+  PRIZMKIT_EFFORT       AI reasoning effort (low|medium|high|xhigh|max; max is Claude Code only)
   VERBOSE               Set to 1 for verbose AI CLI output
   HEARTBEAT_INTERVAL    Heartbeat log interval in seconds (default: 30)
   STALE_KILL_THRESHOLD  Auto-kill after N seconds no progress (default: 900)
@@ -117,6 +122,7 @@ Examples:
   ./run-recovery.sh run --dry-run       # Generate prompt, don't execute
   ./run-recovery.sh run --yes           # Skip confirmation prompt
   ./run-recovery.sh run --model claude-opus-4.6
+  PRIZMKIT_EFFORT=high ./run-recovery.sh   # Use high reasoning effort
 HELP
 }
 

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

@@ -52,6 +52,7 @@ LOG_RETENTION_DAYS=${LOG_RETENTION_DAYS:-14}
 LOG_MAX_TOTAL_MB=${LOG_MAX_TOTAL_MB:-1024}
 VERBOSE=${VERBOSE:-0}
 MODEL=${MODEL:-""}
+PRIZMKIT_EFFORT=${PRIZMKIT_EFFORT:-""}
 DEV_BRANCH=${DEV_BRANCH:-""}
 AUTO_PUSH=${AUTO_PUSH:-0}
 STOP_ON_FAILURE=${STOP_ON_FAILURE:-0}
@@ -61,6 +62,9 @@ ENABLE_DEPLOY=${ENABLE_DEPLOY:-0}
 # Source shared common helpers (CLI/platform detection + logs + deps)
 prizm_detect_cli_and_platform
 
+# Validate PRIZMKIT_EFFORT early (fail-fast before any sessions are spawned)
+prizm_validate_effort
+
 # Source shared heartbeat library
 source "$SCRIPT_DIR/lib/heartbeat.sh"
 
@@ -1232,6 +1236,7 @@ show_help() {
     echo "  MAX_RETRIES              Max retries per refactor (default: 3)"
     echo "  SESSION_TIMEOUT          Session timeout in seconds (default: 0 = no limit)"
     echo "  MODEL                    Default AI model (overridden by per-refactor model in refactor 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)"
     echo "  ENABLE_CRITIC            Enable/disable critic: true|false|1|0 (overridden by --critic/--no-critic)"
     echo "  AI_CLI                   AI CLI command name (auto-detected: cbc, claude, or codex)"