Version 1.35a

  - Changed `afl-multicore` to prefix filenames with their respective job
    number handed over to `afl-fuzz`'s `-f` switch. Previously the job
    number was simply appended causing trouble for some fuzzing targets.
    (Contributed by Krishna Ram Prakash R and @YourButterfly.)
  - Changed `afl-multicore` to also set environment variables specified in the
    `afl-multicore` configuration file in case non-interactive mode is used.
    (Contributed by Philipp Bartsch.)
  - Updated `afl-sync` to not use the `-c` flag when invoking `rsync` to
    avoid hashing of large corpora. (Proposed by Denis Kasak.)
  - Fixed an `UnboundLocalError` in `afl_stats.py` when attempting to parse
    an empty stats file. (Contributed by Vincent Ulitzsch.)

Version 1.34a

  - Added `--chmod`, `--chown` rsync flags to afl-sync (contributed by
    Denis Kasak).
  - Improved bug fix for #34 in afl-multicore (suggested by Bhargava
    Shastry).
  - Bug in afl_collect.py fixed that prevented other tools from starting
    if gdb executable is not present (reported by Henri Salo).
  - Added support for `-m none` to `afl-minimize`. Thus an infinite memory
    limit can be passed to `afl-cmin` and `afl-tmin`. (Contributed by
    Vincent Ulitzsch.)
  - Fixed `afl-multicore` output to display master PID when using
    non-interactive mode.
  - Timestamp field added to `afl-collect` crash sample database.
    WARNING: This breaks compatibility with existing old database files
    in a way that appending to old databases is no longer possible w/o
    updating the database schema to include the timestamp field.

Version 1.33a

  - Added `--cmin-qemu`, `--tmin-qemu` options for QEMU mode support
    to afl-minimize (suggested by Isaac).
  - Made processing timeout for sample verification phase configurable
    in afl-collect.
  - Usage of python 3 virtualenv explicitly documented in README
    (Henri Salo).
  - Added quotes around sample filenames used in generated gdb script
    to keep gdb happy with fancy sample filenames.
  - Updated afl-collect to automatically detect exploitable.py location
    (contributed by Jurriaan Bremer).

Version 1.32a

  - SQLite db connector improved. Gives a huge afl-collect speed-up
    when using a database.
  - Changed afl-collect to print database related outputs only if
    a database is used.
  - Updated afl-stats to be compatible with afl-fuzz >= 2.32b (older
    afl versions will not work anymore).
  - afl-stats now optionally dumps fuzzer stats into a database.
  - Tweeting stats to twitter is now optional in afl-stats.
  - Fixed minor bug #34 in afl-multicore (reported by Bhargava Shastry).
  - Implemented option in afl-multicore to run arbitrary fuzzer instead
    of default afl-fuzz (#35, suggested by Bhargava Shastry).
  - Implemented multi-master mode support in afl-multicore (#36).
  - Bug affecting job counts in afl-multicore fixed (reported by Henri
    Salo).

Version 1.31a

  - Selective resumes added to afl-multicore.
  - Automatic calculation of delay values for afl-multicore startup
    implemented (use `-s auto`).
  - afl-multicore updated to not use a hard-coded path to the
    afl-fuzz binary.

Latest single-threading release.

Version 1.30a

  - Parsing of slightly different modified 'fuzzer_stats' file fixed
    in afl-stats.
  - Delayed startup added to afl-multicore.
  - Fixed a bug in afl-sync that caused some directories to not be pulled
    from the remote location when a session name was specified.
  - Added afl's .cur_input to the rsync exclude list in afl-sync.

Version 1.29a

  - afl-collect updated to not use a hard-coded path to the gdb binary
    (suggested by Martin Lindhe).
  - Fixed #30: CPU affinity settings removed from afl-multicore. (The
    option for explicitly setting CPU affinity in afl was dropped in
    afl-2.17b.)

Version 1.28a

  - afl-cron for periodic task execution added.
  - Main execution loop removed from afl-stats. afl-cron may be used for
    repeated executions of afl-stats!
  - Changed afl-utils to use JSON config files.

Version 1.27a

  - Basic version of afl-sync added.

Version 1.26a

  - afl-minimize now supports reseeding original afl queues with an optimized
    corpus.
  - Typo in test case setup method declaration fixed.
  - More test cases added.

Version 1.25a

  - Added CPU affinity option to afl-multicore.

Version 1.24a

  - afl-minimize now takes timeout and memory limit arguments that are passed
    to afl-cmin and afl-tmin.
  - Updated afl-collect to automatically detect whether operating on a single
    instance output directory or a multi instance synchronisation dir.
  - Fixed a minor bug in afl-collect that occurred when generating output sample
    names.
  - Scrots updated.

Version 1.23a

  - Minor bug-fixes for SampleIndex.
  - Updated afl-multicore to assure that newly created screen windows operate
    on the same directory afl-multicore was started from.
  - Bug fixed, that prevented use of afl -f file argument when running multiple
    afl instances. For this purpose '%%' was introduced to be used when
    referencing the desired file in the target's command line option (check
    README for details).
  - Refactored code to increase testability.
  - Remaining test cases completed and new tests were added.

Version 1.22a

  - Typo in afl-vcrash code fixed (by Emanuele Cozzi).
  - Added flag for configurable sample processing timeouts for afl-vcrash (by
    Emanuele Cozzi).
  - More tests added.

Version 1.21a

  - Ineffective subprocess timeout during crash verification fixed for afl-collect
    and afl-vcrash.
  - Added timeout handling to afl-collect, afl-minimize and afl-vcrash to avoid
    infinite/excessive blocking during sample processing.
  - Unreliable check_screen() fixed in afl-multicore.
  - Added a few test cases for afl-multicore.

Version 1.20a

  - afl-multicore docs updated to reflect latest changes (by Mark Janssen).
  - Fixed afl-multicore session check bug, that prevented adding additional
    afl instances in normal mode (spotted by Mark Janssen).
  - afl-multicore now properly updates its session file when adding additional
    instances. afl-multikill was adapted accordingly.
  - Simple interactive test mode added to afl-multicore.

Version 1.19a

  - afl-multicore revamped. Create config file for your target and desired afl
    options. Easily start and resume configured fuzzer instances. New feature:
    Subsequently add new instances using the 'add' command.
  - afl-multikill updated to terminate fuzzing sessions by process group id
    instead of individual PIDs and using SIGTERM instead of SIGKILL (by Mark
    Janssen).
  - Auto-installation of my hacked version of exploitable added (make sure to
    source exploitable.py as indicated during setup!).

Version 1.18a

  - Bug fixed that caused afl-stats to crash when monitoring more than one fuzzer
    output directory.
  - Added option to afl-collect that simplifies crash sample file names keeping only
    the originating fuzzer name and sample ID (by Martin Gallo).

Version 1.17a

  - afl-minimize will skip file collection if collection dir exists and is not empty.
    This way you can run the automated afl-cmin and afl-tmin invocations directly on
    any directory containing fuzzing samples.
  - All tools' outputs have been colorized.
  - https://github.com/rc0r/exploitable has been updated to avoid crashing when
    multiple inferiors have been detected. Instead an UNKNOWN classification with
    an according message is generated. Be sure to update exploitable to increase
    afl-collect stability!