App-karr

 view release on metacpan or  search on metacpan

.claude/skills/karr-foundation-cli/SKILL.md  view on Meta::CPAN

claude_bin: claude        # binary for claude: true (default: claude)
claude_max_turns: 30      # --max-turns for claude: true (default: 30)
claude_permission_mode: bypassPermissions   # (default: bypassPermissions)
prompt: >-                # agent instruction, exposed to the command as $PROMPT
  Use the karr-coordinator skill: pick the next actionable task and move it.
# command: claude -p "$PROMPT"   # explicit command; wins over claude: true
on_idle: skip             # 'skip' (default) | 'always-run'
drain: true               # loop until drained (default) | false for single run
max_runtime: 1800         # seconds: per-command SIGKILL (0 = no timeout)
max_attempts: 2           # stalls on one task before auto-block (default: 2)
max_iterations: 50        # hard cap on drain iterations / drain budget (default: 50)
cooldown_base: 1          # cooldown minutes at level 0 (default: 1)
cooldown_max: 64          # cooldown ceiling in minutes (default: 64)
error_patterns:           # extra case-insensitive substrings → common-error
  - my custom api error
```

`claude`, `claude_bin`, the `claude_*` knobs, `command` and `prompt` may also be
set globally in `config.yml` (`default_command` / `default_prompt`); the per-repo
`.karr` value wins.

Changes  view on Meta::CPAN

      `karr` CLI.
    - karr-foundation: drain each board instead of a single run — invoke
      the agent command repeatedly until no actionable task (non-terminal
      and unblocked) remains. A task the agent claims but never moves is
      auto-blocked after `max_attempts` stalls (default 2) so the drain
      always terminates; the agent's own `--block` reason still wins.
      Observable common errors (non-zero/timeout exit, or a log match
      against rate-limit/auth/network/5xx patterns, extensible via
      `error_patterns`) never penalize a task and instead trigger an
      exponential per-repo cooldown (1, 2, 4, … minutes, capped). New
      `.karr` keys: `drain`, `max_attempts`, `max_iterations`,
      `cooldown_base`, `cooldown_max`, `error_patterns`.
    - cpanfile: require Git::Native 0.003 and Git::Libgit2 0.004.
    - Fix `karr context` / `karr context --json` crashing with
      "Can't locate object method 'strftime' via package 'Sun May ...'":
      Cmd::Context now `use Time::Piece`, so `gmtime` returns a
      Time::Piece object instead of a plain string. Added t/07-context.t
      covering the plain, --json, and recently-completed cutoff paths.
    - Fix `karr config show` (and get/set) crashing with
      "Can't locate object method 'board_dir'": Cmd::Config now builds
      its config via `$self->store->effective_config` and persists with

lib/App/karr/Foundation.pm  view on Meta::CPAN

# Drain loop
# ---------------------------------------------------------------------------

# Run the agent repeatedly until the board has no actionable tasks left,
# auto-blocking tasks the agent keeps failing on. Returns
# { outcome => progress|idle|common-error|error, exit => N }.
sub _drain_repo {
  my ( $self, $repo, $karr, $cmd ) = @_;
  my $max_runtime  = $karr->{max_runtime}    // 1800;
  my $max_attempts = $karr->{max_attempts}   // 2;
  my $max_iter     = $karr->{max_iterations} // 50;
  my $drain        = exists $karr->{drain} ? $karr->{drain} : 1;
  my $patterns     = $self->_error_patterns( $karr );

  # Use the resolved command, not $karr->{command}
  $cmd //= $karr->{command};

  my $loop_start = time;
  my $last_exit  = 0;
  my $outcome    = 'idle';
  my $first      = 1;

lib/App/karr/Foundation.pm  view on Meta::CPAN

  claude_bin: claude        # binary for claude: true (default: claude)
  claude_max_turns: 30      # --max-turns for claude: true (default: 30)
  claude_permission_mode: bypassPermissions   # (default: bypassPermissions)
  prompt: >-                # agent instruction, exposed as $PROMPT
    Use the karr-coordinator skill: pick the next actionable task and move it.
  command: claude -p "$PROMPT"   # explicit command; wins over claude: true
  on_idle: skip             # 'skip' (default) | 'always-run'
  max_runtime: 1800         # seconds: per-command SIGKILL (0 = no limit)
  drain: true               # loop until drained (default) | false for single run
  max_attempts: 2           # stalls on one task before auto-block (default: 2)
  max_iterations: 50        # hard cap on drain iterations (default: 50)
  cooldown_base: 1          # cooldown minutes at level 0 (default: 1)
  cooldown_max: 64          # cooldown ceiling in minutes (default: 64)
  error_patterns:           # extra case-insensitive substrings → common-error
    - my custom api error

C<claude>, C<claude_bin>, C<claude_max_turns>, C<claude_permission_mode>,
C<command> and C<prompt>/C<default_prompt> may also be set globally in the
config file; the per-repo F<.karr> value wins.

B<Coordinator and overview.> Agent execution is opt-in — a board runs an agent