diff --git a/.clang-tidy b/.clang-tidy
index 5d737ee3f..508250276 100644
--- a/.clang-tidy
+++ b/.clang-tidy
@@ -1,5 +1,5 @@
 ---
-Checks:          'clang-diagnostic-*,clang-analyzer-*,cert-*,performance-*,portability-*'
+Checks:          'clang-diagnostic-*,clang-analyzer-*,cert-*,performance-*,portability-*,modernize-use-auto'
 WarningsAsErrors: ''
 HeaderFilterRegex: ''
 AnalyzeTemporaryDtors: false
@@ -9,5 +9,7 @@ CheckOptions:
     value:           'L;LL;LU;LLU'
   - key:             cert-oop54-cpp.WarnOnlyIfThisHasSuspiciousField
     value:           '0'
+  - key:             modernize-use-auto.RemoveStars
+    value:           '1'
 ...
 
diff --git a/src/builtin_printf.cpp b/src/builtin_printf.cpp
index bda9d974b..8b46a20ba 100644
--- a/src/builtin_printf.cpp
+++ b/src/builtin_printf.cpp
@@ -488,7 +488,7 @@ void builtin_printf_state_t::print_direc(const wchar_t *start, size_t length, wc
     switch (conversion) {
         case L'd':
         case L'i': {
-            intmax_t arg = string_to_scalar_type<intmax_t>(argument, this);
+            auto arg = string_to_scalar_type<intmax_t>(argument, this);
             if (!have_field_width) {
                 if (!have_precision)
                     this->append_format_output(fmt.c_str(), arg);
@@ -506,7 +506,7 @@ void builtin_printf_state_t::print_direc(const wchar_t *start, size_t length, wc
         case L'u':
         case L'x':
         case L'X': {
-            uintmax_t arg = string_to_scalar_type<uintmax_t>(argument, this);
+            auto arg = string_to_scalar_type<uintmax_t>(argument, this);
             if (!have_field_width) {
                 if (!have_precision)
                     this->append_format_output(fmt.c_str(), arg);
@@ -528,7 +528,7 @@ void builtin_printf_state_t::print_direc(const wchar_t *start, size_t length, wc
         case L'F':
         case L'g':
         case L'G': {
-            long double arg = string_to_scalar_type<long double>(argument, this);
+            auto arg = string_to_scalar_type<long double>(argument, this);
             if (!have_field_width) {
                 if (!have_precision) {
                     this->append_format_output(fmt.c_str(), arg);
@@ -579,7 +579,7 @@ void builtin_printf_state_t::print_direc(const wchar_t *start, size_t length, wc
 static inline void modify_allowed_format_specifiers(bool ok[UCHAR_MAX + 1], const char *str,
                                                     bool flag) {
     for (const char *c = str; *c != '\0'; c++) {
-        unsigned char idx = static_cast<unsigned char>(*c);
+        auto idx = static_cast<unsigned char>(*c);
         ok[idx] = flag;
     }
 }
@@ -654,7 +654,7 @@ int builtin_printf_state_t::print_formatted(const wchar_t *format, int argc, wch
                     ++f;
                     ++direc_length;
                     if (argc > 0) {
-                        intmax_t width = string_to_scalar_type<intmax_t>(*argv, this);
+                        auto width = string_to_scalar_type<intmax_t>(*argv, this);
                         if (INT_MIN <= width && width <= INT_MAX)
                             field_width = static_cast<int>(width);
                         else
@@ -679,7 +679,7 @@ int builtin_printf_state_t::print_formatted(const wchar_t *format, int argc, wch
                         ++f;
                         ++direc_length;
                         if (argc > 0) {
-                            intmax_t prec = string_to_scalar_type<intmax_t>(*argv, this);
+                            auto prec = string_to_scalar_type<intmax_t>(*argv, this);
                             if (prec < 0) {
                                 // A negative precision is taken as if the precision were omitted,
                                 // so -1 is safe here even if prec < INT_MIN.
diff --git a/src/builtin_status.cpp b/src/builtin_status.cpp
index 012a18f54..e35de167c 100644
--- a/src/builtin_status.cpp
+++ b/src/builtin_status.cpp
@@ -350,7 +350,7 @@ int builtin_status(parser_t &parser, io_streams_t &streams, wchar_t **argv) {
                 streams.err.append_format(BUILTIN_ERR_ARG_COUNT2, cmd, subcmd_str, 1, args.size());
                 return STATUS_INVALID_ARGS;
             }
-            const auto *metadata = features_t::metadata_for(args.front().c_str());
+            auto metadata = features_t::metadata_for(args.front().c_str());
             if (!metadata) {
                 retval = TEST_FEATURE_NOT_RECOGNIZED;
             } else {
diff --git a/src/builtin_string.cpp b/src/builtin_string.cpp
index 517be1f5a..525f63e73 100644
--- a/src/builtin_string.cpp
+++ b/src/builtin_string.cpp
@@ -871,7 +871,7 @@ class pcre2_matcher_t : public string_matcher_t {
         if (opts.invert_match) return true;
 
         // Report any additional matches.
-        for (auto *ovector = pcre2_get_ovector_pointer(regex.match); opts.all; total_matched++) {
+        for (auto ovector = pcre2_get_ovector_pointer(regex.match); opts.all; total_matched++) {
             uint32_t options = 0;
             PCRE2_SIZE offset = ovector[1];  // start at end of previous match
 
@@ -1052,7 +1052,7 @@ bool regex_replacer_t::replace_matches(const wcstring &arg) {
                        (opts.all ? PCRE2_SUBSTITUTE_GLOBAL : 0);
     size_t arglen = arg.length();
     PCRE2_SIZE bufsize = (arglen == 0) ? 16 : 2 * arglen;
-    wchar_t *output = static_cast<wchar_t *>(malloc(sizeof(wchar_t) * bufsize));
+    auto output = static_cast<wchar_t *>(malloc(sizeof(wchar_t) * bufsize));
     int pcre2_rc;
     PCRE2_SIZE outlen = bufsize;
 
@@ -1071,8 +1071,7 @@ bool regex_replacer_t::replace_matches(const wcstring &arg) {
             done = true;
         } else {
             bufsize = outlen;
-            wchar_t *new_output =
-                static_cast<wchar_t *>(realloc(output, sizeof(wchar_t) * bufsize));
+            auto new_output = static_cast<wchar_t *>(realloc(output, sizeof(wchar_t) * bufsize));
             if (new_output) output = new_output;
         }
     }
@@ -1307,7 +1306,7 @@ static int string_sub(parser_t &parser, io_streams_t &streams, int argc, wchar_t
             pos = static_cast<size_type>(opts.start - 1);
         } else if (opts.start < 0) {
             assert(opts.start != LONG_MIN);  // checked above
-            size_type n = static_cast<size_type>(-opts.start);
+            auto n = static_cast<size_type>(-opts.start);
             pos = n > s->length() ? 0 : s->length() - n;
         }
 
diff --git a/src/color.cpp b/src/color.cpp
index bfb2a52c8..c80195c49 100644
--- a/src/color.cpp
+++ b/src/color.cpp
@@ -87,15 +87,15 @@ static int parse_hex_digit(wchar_t x) {
 }
 
 static unsigned long squared_difference(long p1, long p2) {
-    unsigned long diff = static_cast<unsigned long>(labs(p1 - p2));
+    auto diff = static_cast<unsigned long>(labs(p1 - p2));
     return diff * diff;
 }
 
 static unsigned char convert_color(const unsigned char rgb[3], const uint32_t *colors,
                                    size_t color_count) {
     long r = rgb[0], g = rgb[1], b = rgb[2];
-    unsigned long best_distance = static_cast<unsigned long>(-1);
-    unsigned char best_index = static_cast<unsigned char>(-1);
+    auto best_distance = static_cast<unsigned long>(-1);
+    auto best_index = static_cast<unsigned char>(-1);
     for (size_t idx = 0; idx < color_count; idx++) {
         uint32_t color = colors[idx];
         long test_r = (color >> 16) & 0xFF, test_g = (color >> 8) & 0xFF,
diff --git a/src/common.cpp b/src/common.cpp
index 0f33f7364..8271d4d94 100644
--- a/src/common.cpp
+++ b/src/common.cpp
@@ -361,7 +361,7 @@ char *wcs2str(const wchar_t *in, size_t len) {
     }
 
     // Here we probably allocate a buffer probably much larger than necessary.
-    char *out = static_cast<char *>(malloc(MAX_UTF8_BYTES * len + 1));
+    auto out = static_cast<char *>(malloc(MAX_UTF8_BYTES * len + 1));
     assert(out);
     // Instead of returning the return value of wcs2str_internal, return `out` directly.
     // This eliminates false warnings in coverity about resource leaks.
@@ -1210,7 +1210,7 @@ wcstring debug_escape(const wcstring &in) {
     wcstring result;
     result.reserve(in.size());
     for (wchar_t wc : in) {
-        uint32_t c = static_cast<uint32_t>(wc);
+        auto c = static_cast<uint32_t>(wc);
         if (c <= 127 && isprint(c)) {
             result.push_back(wc);
             continue;
@@ -2148,7 +2148,7 @@ void assert_is_background_thread(const char *who) {
 }
 
 void assert_is_locked(void *vmutex, const char *who, const char *caller) {
-    std::mutex *mutex = static_cast<std::mutex *>(vmutex);
+    auto mutex = static_cast<std::mutex *>(vmutex);
 
     // Note that std::mutex.try_lock() is allowed to return false when the mutex isn't
     // actually locked; fortunately we are checking the opposite so we're safe.
diff --git a/src/complete.cpp b/src/complete.cpp
index 65e98b5de..bbc5a59e1 100644
--- a/src/complete.cpp
+++ b/src/complete.cpp
@@ -422,7 +422,7 @@ bool completer_t::condition_test(const wcstring &condition) {
 
     ASSERT_IS_MAIN_THREAD();
     bool test_res;
-    condition_cache_t::iterator cached_entry = condition_cache.find(condition);
+    auto cached_entry = condition_cache.find(condition);
     if (cached_entry == condition_cache.end()) {
         // Compute new value and reinsert it.
         test_res =
@@ -476,7 +476,7 @@ void complete_add(const wchar_t *cmd, bool cmd_is_path, const wcstring &option,
 /// option strings. Returns true if it is now empty and should be deleted, false if it's not empty.
 /// Must be called while locked.
 bool completion_entry_t::remove_option(const wcstring &option, complete_option_type_t type) {
-    option_list_t::iterator iter = this->options.begin();
+    auto iter = this->options.begin();
     while (iter != this->options.end()) {
         if (iter->option == option && iter->type == type) {
             iter = this->options.erase(iter);
@@ -493,10 +493,10 @@ void complete_remove(const wcstring &cmd, bool cmd_is_path, const wcstring &opti
     auto completion_set = s_completion_set.acquire();
 
     completion_entry_t tmp_entry(cmd, cmd_is_path);
-    completion_entry_set_t::iterator iter = completion_set->find(tmp_entry);
+    auto iter = completion_set->find(tmp_entry);
     if (iter != completion_set->end()) {
         // const_cast: See SET_ELEMENTS_ARE_IMMUTABLE.
-        completion_entry_t &entry = const_cast<completion_entry_t &>(*iter);
+        auto &entry = const_cast<completion_entry_t &>(*iter);
 
         bool delete_it = entry.remove_option(option, type);
         if (delete_it) {
@@ -1788,7 +1788,7 @@ bool complete_remove_wrapper(const wcstring &command, const wcstring &target_to_
     auto locked_map = wrapper_map.acquire();
     wrapper_map_t &wraps = *locked_map;
     bool result = false;
-    wrapper_map_t::iterator current_targets_iter = wraps.find(command);
+    auto current_targets_iter = wraps.find(command);
     if (current_targets_iter != wraps.end()) {
         wcstring_list_t *targets = &current_targets_iter->second;
         auto where = std::find(targets->begin(), targets->end(), target_to_remove);
diff --git a/src/env.cpp b/src/env.cpp
index cdefd3161..d7c526403 100644
--- a/src/env.cpp
+++ b/src/env.cpp
@@ -107,7 +107,7 @@ const electric_var_t *electric_var_t::for_name(const wcstring &name) {
 
 /// Check if a variable may not be set using the set command.
 static bool is_read_only(const wcstring &key) {
-    if (const auto *ev = electric_var_t::for_name(key)) {
+    if (auto ev = electric_var_t::for_name(key)) {
         return ev->readonly();
     }
     // Hack.
diff --git a/src/env_universal_common.cpp b/src/env_universal_common.cpp
index 507f2301f..3fab6cabc 100644
--- a/src/env_universal_common.cpp
+++ b/src/env_universal_common.cpp
@@ -253,13 +253,13 @@ env_universal_t::env_universal_t(wcstring path)
     : narrow_vars_path(wcs2string(path)), explicit_vars_path(std::move(path)) {}
 
 maybe_t<env_var_t> env_universal_t::get(const wcstring &name) const {
-    var_table_t::const_iterator where = vars.find(name);
+    auto where = vars.find(name);
     if (where != vars.end()) return where->second;
     return none();
 }
 
 maybe_t<env_var_t::env_var_flags_t> env_universal_t::get_flags(const wcstring &name) const {
-    var_table_t::const_iterator where = vars.find(name);
+    auto where = vars.find(name);
     if (where != vars.end()) {
         return where->second.get_flags();
     }
@@ -360,7 +360,7 @@ void env_universal_t::generate_callbacks_and_update_exports(const var_table_t &n
 void env_universal_t::acquire_variables(var_table_t &vars_to_acquire) {
     // Copy modified values from existing vars to vars_to_acquire.
     for (const auto &key : this->modified) {
-        var_table_t::iterator src_iter = this->vars.find(key);
+        auto src_iter = this->vars.find(key);
         if (src_iter == this->vars.end()) {
             /* The value has been deleted. */
             vars_to_acquire.erase(key);
diff --git a/src/exec.cpp b/src/exec.cpp
index b575e7aa5..c8d58a5d7 100644
--- a/src/exec.cpp
+++ b/src/exec.cpp
@@ -93,8 +93,8 @@ static void safe_launch_process(process_t *p, const char *actual_cmd, const char
     int err;
 
     // This function never returns, so we take certain liberties with constness.
-    char *const *envv = const_cast<char *const *>(cenvv);
-    char *const *argv = const_cast<char *const *>(cargv);
+    const auto envv = const_cast<char *const *>(cenvv);
+    const auto argv = const_cast<char *const *>(cargv);
 
     execve(actual_cmd, argv, envv);
     err = errno;
@@ -1062,8 +1062,7 @@ static void populate_subshell_output(wcstring_list_t *lst, const io_buffer_t &bu
             const char *cursor = begin;
             while (cursor < end) {
                 // Look for the next separator.
-                const char *stop =
-                    static_cast<const char *>(std::memchr(cursor, '\n', end - cursor));
+                auto stop = static_cast<const char *>(std::memchr(cursor, '\n', end - cursor));
                 const bool hit_separator = (stop != nullptr);
                 if (!hit_separator) {
                     // If it's not found, just use the end.
diff --git a/src/fallback.cpp b/src/fallback.cpp
index 6b7ff3d62..30742136c 100644
--- a/src/fallback.cpp
+++ b/src/fallback.cpp
@@ -77,7 +77,7 @@ int fish_mkstemp_cloexec(char *name_template) {
 // cppcheck-suppress unusedFunction
 [[gnu::unused]] static wchar_t *wcsdup_fallback(const wchar_t *in) {
     size_t len = std::wcslen(in);
-    wchar_t *out = static_cast<wchar_t *>(malloc(sizeof(wchar_t) * (len + 1)));
+    auto out = static_cast<wchar_t *>(malloc(sizeof(wchar_t) * (len + 1)));
     if (out == nullptr) {
         return nullptr;
     }
@@ -162,7 +162,7 @@ int wcsncasecmp(const wchar_t *a, const wchar_t *b, size_t n) {
 
 #ifndef HAVE_WCSNDUP
 wchar_t *wcsndup(const wchar_t *in, size_t c) {
-    wchar_t *res = static_cast<wchar_t *>(malloc(sizeof(wchar_t) * (c + 1)));
+    auto res = static_cast<wchar_t *>(malloc(sizeof(wchar_t) * (c + 1)));
     if (res == nullptr) {
         return nullptr;
     }
diff --git a/src/fish.cpp b/src/fish.cpp
index cbebe02cf..c75c52a1c 100644
--- a/src/fish.cpp
+++ b/src/fish.cpp
@@ -336,12 +336,12 @@ static int fish_parse_opt(int argc, char **argv, fish_cmd_opts_t *opts) {
                 auto cats = get_flog_categories();
                 // Compute width of longest name.
                 int name_width = 0;
-                for (const auto *cat : cats) {
+                for (auto cat : cats) {
                     name_width = std::max(name_width, static_cast<int>(wcslen(cat->name)));
                 }
                 // A little extra space.
                 name_width += 2;
-                for (const auto *cat : cats) {
+                for (auto cat : cats) {
                     // Negating the name width left-justifies.
                     printf("%*ls %ls\n", -name_width, cat->name, _(cat->description));
                 }
diff --git a/src/fish_tests.cpp b/src/fish_tests.cpp
index a197dfa48..8312d12d5 100644
--- a/src/fish_tests.cpp
+++ b/src/fish_tests.cpp
@@ -810,8 +810,8 @@ static void test_fd_monitor() {
     item_oneshot.always_exit = true;
     {
         fd_monitor_t monitor;
-        for (auto *item : {&item_never, &item_hugetimeout, &item0_timeout, &item42_timeout,
-                           &item42_nottimeout, &item42_thenclose, &item_oneshot}) {
+        for (auto item : {&item_never, &item_hugetimeout, &item0_timeout, &item42_timeout,
+                          &item42_nottimeout, &item42_thenclose, &item_oneshot}) {
             monitor.add(std::move(item->item));
         }
         item42_timeout.write42();
diff --git a/src/function.cpp b/src/function.cpp
index 56e5fb2c5..e841d55fc 100644
--- a/src/function.cpp
+++ b/src/function.cpp
@@ -80,7 +80,7 @@ static owning_lock<function_set_t> function_set;
 bool function_set_t::allow_autoload(const wcstring &name) const {
     // Prohibit autoloading if we have a non-autoload (explicit) function, or if the function is
     // tombstoned.
-    const auto *info = get_info(name);
+    auto info = get_info(name);
     bool has_explicit_func = info && !info->is_autoload;
     bool is_tombstoned = autoload_tombstones.count(name) > 0;
     return !has_explicit_func && !is_tombstoned;
@@ -175,7 +175,7 @@ void function_add(wcstring name, wcstring description, function_properties_ref_t
 std::shared_ptr<const function_properties_t> function_get_properties(const wcstring &name) {
     if (parser_keywords_is_reserved(name)) return nullptr;
     auto funcset = function_set.acquire();
-    if (const auto *info = funcset->get_info(name)) {
+    if (auto info = funcset->get_info(name)) {
         return info->props;
     }
     return nullptr;
diff --git a/src/history.cpp b/src/history.cpp
index d783035d0..980fc50a5 100644
--- a/src/history.cpp
+++ b/src/history.cpp
@@ -466,8 +466,7 @@ void history_impl_t::set_valid_file_paths(const wcstring_list_t &valid_file_path
     }
 
     // Look for an item with the given identifier. It is likely to be at the end of new_items.
-    for (history_item_list_t::reverse_iterator iter = new_items.rbegin(); iter != new_items.rend();
-         ++iter) {
+    for (auto iter = new_items.rbegin(); iter != new_items.rend(); ++iter) {
         if (iter->identifier == ident) {  // found it
             iter->required_paths = valid_file_paths;
             break;
@@ -603,7 +602,7 @@ void history_impl_t::load_old_if_needed() {
 
 bool history_search_t::go_backwards() {
     // Backwards means increasing our index.
-    const size_t max_index = static_cast<size_t>(-1);
+    const auto max_index = static_cast<size_t>(-1);
 
     if (current_index_ == max_index) return false;
 
diff --git a/src/history_file.cpp b/src/history_file.cpp
index 453c96dac..66cc2f548 100644
--- a/src/history_file.cpp
+++ b/src/history_file.cpp
@@ -29,7 +29,7 @@ static bool should_mmap(int fd) {
 // Return true on success, false on failure.
 static bool read_from_fd(int fd, void *address, size_t len) {
     size_t remaining = len;
-    char *ptr = static_cast<char *>(address);
+    auto ptr = static_cast<char *>(address);
     while (remaining > 0) {
         ssize_t amt = read(fd, ptr, remaining);
         if (amt < 0) {
@@ -162,7 +162,7 @@ history_item_t history_file_contents_t::decode_item(size_t offset) const {
 }
 
 maybe_t<size_t> history_file_contents_t::offset_of_next_item(size_t *cursor, time_t cutoff) {
-    size_t offset = size_t(-1);
+    auto offset = size_t(-1);
     switch (this->type()) {
         case history_type_fish_2_0:
             offset = offset_of_next_item_fish_2_0(*this, cursor, cutoff);
@@ -183,7 +183,7 @@ static size_t read_line(const char *base, size_t cursor, size_t len, std::string
     // Locate the newline.
     assert(cursor <= len);
     const char *start = base + cursor;
-    const char *a_newline = static_cast<const char *>(std::memchr(start, '\n', len - cursor));
+    auto a_newline = static_cast<const char *>(std::memchr(start, '\n', len - cursor));
     if (a_newline != nullptr) {  // we found a newline
         result.assign(start, a_newline - start);
         // Return the amount to advance the cursor; skip over the newline.
@@ -342,7 +342,7 @@ static const char *next_line(const char *start, const char *end) {
 static size_t offset_of_next_item_fish_2_0(const history_file_contents_t &contents,
                                            size_t *inout_cursor, time_t cutoff_timestamp) {
     size_t cursor = *inout_cursor;
-    size_t result = size_t(-1);
+    auto result = size_t(-1);
     const size_t length = contents.length();
     const char *const begin = contents.begin();
     const char *const end = contents.end();
@@ -350,8 +350,7 @@ static size_t offset_of_next_item_fish_2_0(const history_file_contents_t &conten
         const char *line_start = contents.address_at(cursor);
 
         // Advance the cursor to the next line.
-        const char *a_newline =
-            static_cast<const char *>(std::memchr(line_start, '\n', length - cursor));
+        auto a_newline = static_cast<const char *>(std::memchr(line_start, '\n', length - cursor));
         if (a_newline == nullptr) break;
 
         // Advance the cursor past this line. +1 is for the newline.
@@ -509,7 +508,7 @@ static history_item_t decode_item_fish_1_x(const char *begin, size_t length) {
                 while (*time_string && !iswdigit(*time_string)) time_string++;
 
                 if (*time_string) {
-                    time_t tm = static_cast<time_t>(fish_wcstol(time_string));
+                    auto tm = static_cast<time_t>(fish_wcstol(time_string));
                     if (!errno && tm >= 0) {
                         timestamp = tm;
                     }
diff --git a/src/io.cpp b/src/io.cpp
index 42fabfd46..070dd2687 100644
--- a/src/io.cpp
+++ b/src/io.cpp
@@ -31,7 +31,7 @@
 /// Provide the fd monitor used for background fillthread operations.
 static fd_monitor_t &fd_monitor() {
     // Deliberately leaked to avoid shutdown dtors.
-    static fd_monitor_t *fdm = new fd_monitor_t();
+    static auto fdm = new fd_monitor_t();
     return *fdm;
 }
 
@@ -202,7 +202,7 @@ io_buffer_t::~io_buffer_t() {
 
 void io_chain_t::remove(const shared_ptr<const io_data_t> &element) {
     // See if you can guess why std::find doesn't work here.
-    for (io_chain_t::iterator iter = this->begin(); iter != this->end(); ++iter) {
+    for (auto iter = this->begin(); iter != this->end(); ++iter) {
         if (*iter == element) {
             this->erase(iter);
             break;
diff --git a/src/iothread.cpp b/src/iothread.cpp
index b4a8ea651..c7a32f376 100644
--- a/src/iothread.cpp
+++ b/src/iothread.cpp
@@ -466,7 +466,7 @@ using void_func_t = std::function<void(void)>;
 static void *func_invoker(void *param) {
     // Acquire a thread id for this thread.
     (void)thread_id();
-    void_func_t *vf = static_cast<void_func_t *>(param);
+    auto vf = static_cast<void_func_t *>(param);
     (*vf)();
     delete vf;
     return nullptr;
@@ -474,7 +474,7 @@ static void *func_invoker(void *param) {
 
 bool make_detached_pthread(void_func_t &&func) {
     // Copy the function into a heap allocation.
-    void_func_t *vf = new void_func_t(std::move(func));
+    auto vf = new void_func_t(std::move(func));
     if (make_detached_pthread(func_invoker, vf)) {
         return true;
     }
diff --git a/src/null_terminated_array.cpp b/src/null_terminated_array.cpp
index 368584d7f..033f60881 100644
--- a/src/null_terminated_array.cpp
+++ b/src/null_terminated_array.cpp
@@ -22,13 +22,13 @@ static CharT **make_null_terminated_array_helper(
     }
 
     // Now allocate their sum.
-    unsigned char *base =
+    auto base =
         static_cast<unsigned char *>(malloc(pointers_allocation_len + strings_allocation_len));
     if (!base) return nullptr;
 
     // Divvy it up into the pointers and strings.
-    CharT **pointers = reinterpret_cast<CharT **>(base);
-    CharT *strings = reinterpret_cast<CharT *>(base + pointers_allocation_len);
+    auto pointers = reinterpret_cast<CharT **>(base);
+    auto strings = reinterpret_cast<CharT *>(base + pointers_allocation_len);
 
     // Start copying.
     for (size_t i = 0; i < count; i++) {
diff --git a/src/null_terminated_array.h b/src/null_terminated_array.h
index 59950be75..ec8bccdc9 100644
--- a/src/null_terminated_array.h
+++ b/src/null_terminated_array.h
@@ -64,7 +64,7 @@ class null_terminated_array_t {
     /// Convert from a null terminated list to a vector of strings.
     static string_list_t to_list(const CharT *const *arr) {
         string_list_t result;
-        for (const auto *cursor = arr; cursor && *cursor; cursor++) {
+        for (auto cursor = arr; cursor && *cursor; cursor++) {
             result.push_back(*cursor);
         }
         return result;
diff --git a/src/pager.cpp b/src/pager.cpp
index c8a0622b6..672005b70 100644
--- a/src/pager.cpp
+++ b/src/pager.cpp
@@ -84,7 +84,7 @@ static size_t print_max(const wcstring &str, highlight_spec_t color, size_t max,
             // skip non-printable characters
             continue;
         }
-        size_t width_c = size_t(iwidth_c);
+        auto width_c = size_t(iwidth_c);
 
         if (width_c > remaining) break;
 
@@ -139,7 +139,7 @@ line_t pager_t::completion_print_item(const wcstring &prefix, const comp_t *c, s
 
     auto modify_role = [=](highlight_role_t role) -> highlight_role_t {
         using uint_t = typename std::underlying_type<highlight_role_t>::type;
-        uint_t base = static_cast<uint_t>(role);
+        auto base = static_cast<uint_t>(role);
         if (selected) {
             base += static_cast<uint_t>(highlight_role_t::pager_selected_background) -
                     static_cast<uint_t>(highlight_role_t::pager_background);
diff --git a/src/parse_tree.cpp b/src/parse_tree.cpp
index 9b9c745b6..46ee115ab 100644
--- a/src/parse_tree.cpp
+++ b/src/parse_tree.cpp
@@ -453,7 +453,7 @@ class parse_ll_t {
         // it's lowest on the stack)
         const size_t child_start_big = nodes.size();
         assert(child_start_big < NODE_OFFSET_INVALID);
-        node_offset_t child_start = static_cast<node_offset_t>(child_start_big);
+        auto child_start = static_cast<node_offset_t>(child_start_big);
 
         // To avoid constructing multiple nodes, we make a single one that we modify.
         parse_node_t representative_child(token_type_invalid);
@@ -728,7 +728,7 @@ void parse_ll_t::parse_error_unexpected_token(const wchar_t *expected, parse_tok
 
 void parse_ll_t::reset_symbols(enum parse_token_type_t goal) {
     // Add a new goal node, and then reset our symbol list to point at it.
-    node_offset_t where = static_cast<node_offset_t>(nodes.size());
+    auto where = static_cast<node_offset_t>(nodes.size());
     nodes.push_back(parse_node_t(goal));
 
     symbol_stack.clear();
diff --git a/src/proc.cpp b/src/proc.cpp
index 1bd91739c..a050e9e66 100644
--- a/src/proc.cpp
+++ b/src/proc.cpp
@@ -96,7 +96,7 @@ void proc_init() { signal_set_handlers_once(false); }
 
 // Basic thread safe sorted vector of job IDs in use.
 // This is deliberately leaked to avoid dtor ordering issues - see #6539.
-static auto *const locked_consumed_job_ids = new owning_lock<std::vector<job_id_t>>();
+static const auto locked_consumed_job_ids = new owning_lock<std::vector<job_id_t>>();
 
 job_id_t acquire_job_id() {
     auto consumed_job_ids = locked_consumed_job_ids->acquire();
diff --git a/src/reader.cpp b/src/reader.cpp
index a35bd2f51..67014577d 100644
--- a/src/reader.cpp
+++ b/src/reader.cpp
@@ -148,13 +148,13 @@ operation_context_t get_bg_context(const std::shared_ptr<environment_t> &env,
 /// These are deliberately leaked to avoid shutdown dtor registration.
 static debounce_t &debounce_autosuggestions() {
     const long kAutosuggetTimeoutMs = 500;
-    static debounce_t *res = new debounce_t(kAutosuggetTimeoutMs);
+    static auto res = new debounce_t(kAutosuggetTimeoutMs);
     return *res;
 }
 
 static debounce_t &debounce_highlighting() {
     const long kHighlightTimeoutMs = 500;
-    static debounce_t *res = new debounce_t(kHighlightTimeoutMs);
+    static auto res = new debounce_t(kHighlightTimeoutMs);
     return *res;
 }
 
@@ -549,7 +549,7 @@ class reader_data_t : public std::enable_shared_from_this<reader_data_t> {
     }
 
     editable_line_t *active_edit_line() {
-        const auto *cthis = this;
+        auto cthis = reinterpret_cast<const reader_data_t *>(this);
         return const_cast<editable_line_t *>(cthis->active_edit_line());
     }
 
diff --git a/src/tinyexpr.cpp b/src/tinyexpr.cpp
index e134b6cb7..dc1206154 100644
--- a/src/tinyexpr.cpp
+++ b/src/tinyexpr.cpp
@@ -109,7 +109,7 @@ static te_expr *new_expr(const int type, const te_expr *parameters[]) {
     const int arity = get_arity(type);
     const int psize = sizeof(te_expr *) * arity;
     const int size = sizeof(te_expr) + psize;
-    te_expr *ret = static_cast<te_expr *>(malloc(size));
+    auto ret = static_cast<te_expr *>(malloc(size));
     // This sets float to 0, which depends on the implementation.
     // We rely on IEEE-754 floats anyway, so it's okay.
     std::memset(ret, 0, size);
@@ -142,7 +142,7 @@ static constexpr double e() { return M_E; }
 static double fac(double a) { /* simplest version of fac */
     if (a < 0.0) return NAN;
     if (a > UINT_MAX) return INFINITY;
-    unsigned int ua = static_cast<unsigned int>(a);
+    auto ua = static_cast<unsigned int>(a);
     unsigned long int result = 1, i;
     for (i = 1; i <= ua; i++) {
         if (i > ULONG_MAX / result) return INFINITY;
@@ -448,7 +448,7 @@ static te_expr *factor(state *s) {
 
     while (s->type == TOK_INFIX &&
            (s->function == reinterpret_cast<const void *>(static_cast<te_fun2>(pow)))) {
-        te_fun2 t = (te_fun2)s->function;
+        auto t = (te_fun2)s->function;
         next_token(s);
 
         if (insertion) {
@@ -475,7 +475,7 @@ static te_expr *term(state *s) {
            (s->function == reinterpret_cast<const void *>(static_cast<te_fun2>(mul)) ||
             s->function == reinterpret_cast<const void *>(static_cast<te_fun2>(divide)) ||
             s->function == reinterpret_cast<const void *>(static_cast<te_fun2>(fmod)))) {
-        te_fun2 t = (te_fun2)s->function;
+        auto t = (te_fun2)s->function;
         next_token(s);
         ret = NEW_EXPR(TE_FUNCTION2, ret, factor(s));
         ret->function = reinterpret_cast<const void *>(t);
@@ -489,7 +489,7 @@ static te_expr *expr(state *s) {
     te_expr *ret = term(s);
 
     while (s->type == TOK_INFIX && (s->function == add || s->function == sub)) {
-        te_fun2 t = (te_fun2)s->function;
+        auto t = (te_fun2)s->function;
         next_token(s);
         ret = NEW_EXPR(TE_FUNCTION2, ret, term(s));
         ret->function = reinterpret_cast<const void *>(t);
diff --git a/src/utf8.cpp b/src/utf8.cpp
index 1086fd8f1..f52d3100a 100644
--- a/src/utf8.cpp
+++ b/src/utf8.cpp
@@ -67,7 +67,7 @@ bool wchar_to_utf8_string(const std::wstring &str, std::string *result) {
     const wchar_t *input = str.c_str();
     size_t outlen = wchar_to_utf8(input, inlen, nullptr, 0, 0);
     if (outlen > 0) {
-        char *tmp = new char[outlen];
+        auto tmp = new char[outlen];
         size_t outlen2 = wchar_to_utf8(input, inlen, tmp, outlen, 0);
         if (outlen2 > 0) {
             result->assign(tmp, outlen2);
@@ -110,7 +110,7 @@ size_t wchar_to_utf8(const wchar_t *in, size_t insize, char *out, size_t outsize
     } else {
         // Allocate a temporary buffer to hold the input the std::copy performs the size conversion.
         // Note: insize may be 0.
-        utf8_wchar_t *tmp_input = new utf8_wchar_t[insize];
+        auto tmp_input = new utf8_wchar_t[insize];
         if (!safe_copy_wchar_to_utf8_wchar(in, tmp_input, insize)) {
             // Our utf8_wchar_t is UCS-16 and there was an astral character.
             result = 0;