计划执行与检查节点
executing-plans
按既定实施计划完成工作,保持过程检查和反馈。
cli-and-config【效率与协作】Guides labelme CLI startup, GUI session configuration, label and flag sources, YAML Settings, output paths, and display-related recovery.
独立 GitHub 快照 · 自动分类 · 以原作者文档为准
Use this route when the task is to launch labelme, choose annotation-session options, configure labels/flags, control saving or navigation, or debug a Config File and GUI startup.
labelme in the target environment.labelme --help and labelme --version before opening a GUI.references/cli-reference.md for exact behavior.--labels and --flags with comma-separated values or a text file;
use --label-flags for YAML label-specific Shape Flags.--output as a directory for a session, not a .json filename, unless
the behavior is intentionally a single target as described in the CLI ref.--config <file> for a writable Config File or --config '{...}' for
ephemeral session overrides. Inspect a candidate without writing it with
scripts/inspect_labelme_config.py.--validate-label exact is enabled, provide a non-empty label list.references/troubleshooting.md.~/.labelmerc.true/false are booleans; yes/no/on/off are strings.canvas.allow_out_of_bounds_points is opt-in and preserves negative or
beyond-image coordinates for partially visible objects.QSettings state; --reset-config
clears that state and is not a substitute for editing YAML Overrides.references/cli-reference.md for flags and output semantics.references/configuration.md for Settings groups, migration behavior,
YAML examples, and validation constraints.references/troubleshooting.md for install, Qt/display, and malformed
config recovery.scripts/inspect_labelme_config.py --config-yaml '{labels: [cat]}' --show labels
to validate an inline mapping without writing a Config File.Route JSON schema questions to ../annotation-data/SKILL.md, conversion
questions to ../dataset-export/SKILL.md, AI model/prompt questions to
../ai-assisted-annotation/SKILL.md, and source/test changes to
../repo-development/SKILL.md.