lemming-cli 0.9.1__tar.gz → 0.10.1__tar.gz

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (206) hide show
  1. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/.github/workflows/ci.yml +5 -5
  2. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/.github/workflows/publish.yml +3 -3
  3. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/.github/workflows/screenshots.yml +3 -3
  4. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/Dockerfile +2 -2
  5. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/PKG-INFO +85 -62
  6. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/README.md +83 -60
  7. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/SKILL.md +64 -33
  8. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/biome.json +37 -2
  9. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/docs/EVALS.md +3 -0
  10. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/docs/HOOKS.md +20 -19
  11. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/docs/WORKSPACE.md +9 -5
  12. lemming_cli-0.10.1/docs/screenshots/dashboard-desktop.png +0 -0
  13. lemming_cli-0.10.1/docs/screenshots/dashboard-mobile.png +0 -0
  14. lemming_cli-0.10.1/docs/screenshots/task-log-desktop.png +0 -0
  15. lemming_cli-0.10.1/docs/screenshots/task-log-mobile.png +0 -0
  16. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/package-lock.json +260 -599
  17. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/package.json +5 -6
  18. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/pyproject.toml +2 -2
  19. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/api/config.py +7 -21
  20. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/api/config_test.py +2 -0
  21. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/api/conftest.py +1 -1
  22. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/api/files.py +1 -1
  23. lemming_cli-0.10.1/src/lemming/api/loop.py +70 -0
  24. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/api/loop_test.py +2 -0
  25. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/api/main.py +27 -1
  26. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/api/tasks.py +13 -7
  27. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/api/tasks_test.py +61 -0
  28. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/artifact_test.py +10 -0
  29. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/brief_test.py +42 -0
  30. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/artifact.py +3 -1
  31. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/brief.py +12 -5
  32. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/exec_cli.py +111 -25
  33. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/exec_cli_test.py +377 -15
  34. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/goal.py +6 -6
  35. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/goal_test.py +20 -0
  36. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/machine_output_test.py +2 -2
  37. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/main.py +26 -0
  38. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/operations.py +93 -38
  39. lemming_cli-0.10.1/src/lemming/cli/operations_test.py +529 -0
  40. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/progress.py +2 -2
  41. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/stop_test.py +5 -6
  42. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/tasks.py +79 -22
  43. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/tasks_test.py +164 -11
  44. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/readability.py +0 -1
  45. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/roadmap.py +0 -6
  46. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/trial.py +10 -14
  47. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/integration_test.py +38 -33
  48. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/main.py +7 -0
  49. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/models.py +48 -3
  50. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/models_test.py +21 -1
  51. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/orchestrator.py +265 -603
  52. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/orchestrator_test.py +546 -289
  53. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/prompts/hooks/50-readability.md +27 -26
  54. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/prompts/hooks/60-testing.md +12 -12
  55. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/prompts/hooks/70-ux.md +6 -6
  56. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/prompts/hooks/90-roadmap.md +41 -41
  57. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/prompts/taskrunner.md +31 -30
  58. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/prompts.py +18 -8
  59. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/prompts_test.py +22 -13
  60. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/runner.py +35 -8
  61. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/runner_test.py +145 -12
  62. lemming_cli-0.10.1/src/lemming/scope.py +173 -0
  63. lemming_cli-0.10.1/src/lemming/scope_test.py +215 -0
  64. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/snapshots.py +12 -8
  65. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/snapshots_test.py +57 -2
  66. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/tasks/lifecycle.py +140 -257
  67. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/tasks/lifecycle_test.py +118 -114
  68. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/tasks/operations.py +33 -77
  69. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/tasks/operations_test.py +47 -7
  70. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/tasks/progress.py +2 -0
  71. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/tasks/progress_test.py +1 -0
  72. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/tasks/queries.py +23 -42
  73. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/tasks/queries_test.py +3 -3
  74. lemming_cli-0.10.1/src/lemming/tasks/state_machine.py +336 -0
  75. lemming_cli-0.10.1/src/lemming/tasks/state_machine_test.py +165 -0
  76. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/web/dashboard.spec.js +1 -1
  77. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/web/dashboard.test.js +19 -16
  78. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/web/favicon.spec.js +3 -3
  79. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/web/files.html +164 -166
  80. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/web/index.html +127 -104
  81. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/web/index.js +16 -15
  82. lemming_cli-0.10.1/src/lemming/web/logs.html +386 -0
  83. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/web/logs.test.js +8 -6
  84. lemming_cli-0.10.1/src/lemming/web/mancha.js +46 -0
  85. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/web/screenshots.spec.js +1 -1
  86. lemming_cli-0.10.1/uv.lock +534 -0
  87. lemming_cli-0.9.1/.prettierignore +0 -13
  88. lemming_cli-0.9.1/.prettierrc +0 -6
  89. lemming_cli-0.9.1/docs/screenshots/dashboard-desktop.png +0 -0
  90. lemming_cli-0.9.1/docs/screenshots/dashboard-mobile.png +0 -0
  91. lemming_cli-0.9.1/docs/screenshots/task-log-desktop.png +0 -0
  92. lemming_cli-0.9.1/docs/screenshots/task-log-mobile.png +0 -0
  93. lemming_cli-0.9.1/src/lemming/api/loop.py +0 -45
  94. lemming_cli-0.9.1/src/lemming/cli/operations_test.py +0 -205
  95. lemming_cli-0.9.1/src/lemming/scope.py +0 -173
  96. lemming_cli-0.9.1/src/lemming/scope_test.py +0 -125
  97. lemming_cli-0.9.1/src/lemming/web/logs.html +0 -377
  98. lemming_cli-0.9.1/src/lemming/web/mancha.js +0 -46
  99. lemming_cli-0.9.1/uv.lock +0 -493
  100. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/.dockerignore +0 -0
  101. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/.gitignore +0 -0
  102. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/LICENSE +0 -0
  103. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/docker-compose.yml +0 -0
  104. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/playwright.config.js +0 -0
  105. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/__init__.py +0 -0
  106. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/api/__init__.py +0 -0
  107. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/api/auth.py +0 -0
  108. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/api/auth_test.py +0 -0
  109. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/api/context.py +0 -0
  110. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/api/context_test.py +0 -0
  111. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/api/directories.py +0 -0
  112. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/api/directories_test.py +0 -0
  113. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/api/files_test.py +0 -0
  114. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/api/hooks.py +0 -0
  115. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/api/hooks_test.py +0 -0
  116. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/api/logging.py +0 -0
  117. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/api/logging_test.py +0 -0
  118. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/api/main_test.py +0 -0
  119. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/__init__.py +0 -0
  120. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/config.py +0 -0
  121. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/config_test.py +0 -0
  122. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/hooks.py +0 -0
  123. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/hooks_test.py +0 -0
  124. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/main_test.py +0 -0
  125. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/progress_test.py +0 -0
  126. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/readability_cli.py +0 -0
  127. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/readability_cli_test.py +0 -0
  128. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/skill.py +0 -0
  129. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/cli/skill_test.py +0 -0
  130. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/conftest.py +0 -0
  131. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/__init__.py +0 -0
  132. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/__main__.py +0 -0
  133. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/cli.py +0 -0
  134. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/cli_test.py +0 -0
  135. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/container.py +0 -0
  136. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/container_test.py +0 -0
  137. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/fixtures.py +0 -0
  138. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/fixtures_test.py +0 -0
  139. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/harness.py +0 -0
  140. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/harness_test.py +0 -0
  141. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/metrics.py +0 -0
  142. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/metrics_test.py +0 -0
  143. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/readability/base/README.md +0 -0
  144. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/readability/base/calc/__init__.py +0 -0
  145. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/readability/base/calc/legacy.py +0 -0
  146. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/readability/base/calc/ops.py +0 -0
  147. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/readability/base/calc/ops_test.py +0 -0
  148. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/readability/clean/calc/ops.py +0 -0
  149. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/readability/clean/calc/ops_test.py +0 -0
  150. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/readability/dead-code/calc/ops.py +0 -0
  151. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/readability/dead-code/calc/ops_test.py +0 -0
  152. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/readability/duplicated-behavior/calc/ops.py +0 -0
  153. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/readability/duplicated-behavior/calc/ops_test.py +0 -0
  154. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/readability/false-reuse/calc/limits.py +0 -0
  155. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/readability/false-reuse/calc/limits_test.py +0 -0
  156. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/readability/lint-debt/calc/ops.py +0 -0
  157. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/readability/lint-debt/calc/ops_test.py +0 -0
  158. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/roadmap/add-only/README.md +0 -0
  159. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/roadmap/add-only/calc/__init__.py +0 -0
  160. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/roadmap/add-only/calc/ops.py +0 -0
  161. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/roadmap/add-only/calc/ops_test.py +0 -0
  162. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/roadmap/buggy-subtract/README.md +0 -0
  163. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/roadmap/buggy-subtract/calc/__init__.py +0 -0
  164. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/roadmap/buggy-subtract/calc/ops.py +0 -0
  165. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/roadmap/buggy-subtract/calc/ops_test.py +0 -0
  166. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/roadmap/full-ops/README.md +0 -0
  167. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/roadmap/full-ops/calc/__init__.py +0 -0
  168. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/roadmap/full-ops/calc/ops.py +0 -0
  169. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/roadmap/full-ops/calc/ops_test.py +0 -0
  170. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/roadmap/unwired-cli/README.md +0 -0
  171. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/roadmap/unwired-cli/calc/__init__.py +0 -0
  172. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/roadmap/unwired-cli/calc/cli.py +0 -0
  173. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/roadmap/unwired-cli/calc/cli_test.py +0 -0
  174. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/roadmap/unwired-cli/calc/ops.py +0 -0
  175. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/roadmap/unwired-cli/calc/ops_test.py +0 -0
  176. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/task/hidden/stats/summary_hidden_test.py +0 -0
  177. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/task/summary/README.md +0 -0
  178. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/task/summary/stats/__init__.py +0 -0
  179. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/task/summary/stats/summary.py +0 -0
  180. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/projects/task/summary/stats/summary_test.py +0 -0
  181. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/readability_test.py +0 -0
  182. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/report.py +0 -0
  183. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/report_test.py +0 -0
  184. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/roadmap_test.py +0 -0
  185. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/scenarios.py +0 -0
  186. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/scenarios_test.py +0 -0
  187. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/suites.py +0 -0
  188. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/task.py +0 -0
  189. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/task_test.py +0 -0
  190. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/evals/trial_test.py +0 -0
  191. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/hooks.py +0 -0
  192. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/hooks_test.py +0 -0
  193. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/main_test.py +0 -0
  194. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/paths.py +0 -0
  195. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/paths_test.py +0 -0
  196. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/persistence.py +0 -0
  197. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/persistence_test.py +0 -0
  198. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/providers.py +0 -0
  199. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/providers_test.py +0 -0
  200. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/shutdown.py +0 -0
  201. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/shutdown_integration_test.py +0 -0
  202. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/shutdown_test.py +0 -0
  203. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/tasks/__init__.py +0 -0
  204. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/tasks/limits.py +0 -0
  205. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/web/favicon.js +0 -0
  206. {lemming_cli-0.9.1 → lemming_cli-0.10.1}/src/lemming/web/files.spec.js +0 -0
@@ -14,20 +14,20 @@ jobs:
14
14
  runs-on: ubuntu-latest
15
15
 
16
16
  steps:
17
- - uses: actions/checkout@v6
17
+ - uses: actions/checkout@v7
18
18
 
19
19
  - name: Set up Python
20
- uses: astral-sh/setup-uv@v7
20
+ uses: astral-sh/setup-uv@v10.3.3
21
21
  with:
22
- python-version: '3.14'
22
+ python-version: '3.15'
23
23
 
24
24
  - name: Install Python dependencies
25
25
  run: uv sync --locked
26
26
 
27
27
  - name: Set up Node.js
28
- uses: actions/setup-node@v6
28
+ uses: actions/setup-node@v7
29
29
  with:
30
- node-version: '20'
30
+ node-version: '26'
31
31
 
32
32
  # Before the Python tests: they invoke readability, which runs this.
33
33
  - name: Install Node.js dependencies
@@ -12,12 +12,12 @@ jobs:
12
12
  id-token: write
13
13
 
14
14
  steps:
15
- - uses: actions/checkout@v6
15
+ - uses: actions/checkout@v7
16
16
 
17
17
  - name: Set up Python
18
- uses: astral-sh/setup-uv@v7
18
+ uses: astral-sh/setup-uv@v10.3.3
19
19
  with:
20
- python-version: '3.14'
20
+ python-version: '3.15'
21
21
 
22
22
  - name: Build distribution
23
23
  run: uv build
@@ -21,12 +21,12 @@ jobs:
21
21
  runs-on: ubuntu-latest
22
22
 
23
23
  steps:
24
- - uses: actions/checkout@v6
24
+ - uses: actions/checkout@v7
25
25
 
26
26
  - name: Set up Node.js
27
- uses: actions/setup-node@v6
27
+ uses: actions/setup-node@v7
28
28
  with:
29
- node-version: '20'
29
+ node-version: '26'
30
30
 
31
31
  - name: Install Node.js dependencies
32
32
  run: npm ci
@@ -1,4 +1,4 @@
1
- FROM python:3.14-slim
1
+ FROM python:3.15-slim
2
2
 
3
3
  # Install common dev tools
4
4
  RUN apt-get update && apt-get install -y \
@@ -6,7 +6,7 @@ RUN apt-get update && apt-get install -y \
6
6
  && rm -rf /var/lib/apt/lists/*
7
7
 
8
8
  # Install Node.js (LTS) and npm
9
- RUN curl -fsSL https://deb.nodesource.com/setup_24.x | bash - \
9
+ RUN curl -fsSL https://deb.nodesource.com/setup_26.x | bash - \
10
10
  && apt-get install -y nodejs \
11
11
  && rm -rf /var/lib/apt/lists/*
12
12
 
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.5
2
2
  Name: lemming-cli
3
- Version: 0.9.1
3
+ Version: 0.10.1
4
4
  Summary: An autonomous, iterative task runner for AI coding agents.
5
5
  Project-URL: Homepage, https://github.com/owahltinez/lemming
6
6
  Project-URL: Repository, https://github.com/owahltinez/lemming
@@ -14,7 +14,7 @@ Requires-Dist: click>=8.0.0
14
14
  Requires-Dist: fastapi>=0.124.4
15
15
  Requires-Dist: pydantic>=2.12.5
16
16
  Requires-Dist: pyyaml>=6.0.3
17
- Requires-Dist: readability-cli>=0.10.6
17
+ Requires-Dist: readability-cli>=0.11.1
18
18
  Requires-Dist: uvicorn>=0.33.0
19
19
  Description-Content-Type: text/markdown
20
20
 
@@ -95,7 +95,7 @@ Start the autonomous loop.
95
95
  # claude, or codex)
96
96
  lemming run
97
97
 
98
- # Flags passed after -- are sent directly to the underlying runner
98
+ # Arguments after -- are sent directly to the underlying runner
99
99
  lemming run -- --model claude-3-5-sonnet
100
100
  ```
101
101
 
@@ -123,11 +123,11 @@ lemming exec "Fix the flaky test in runner_test.py" --retries 3
123
123
  cat handoff.md | lemming -v exec -f - --runner agy
124
124
  ```
125
125
 
126
- The runner receives no conversation history and cannot be steered mid-run.
127
- Make the handoff self-contained: state the goal, hard constraints, relevant
128
- findings and evidence, approaches ruled out and why, and the definition of done
129
- and verification. The installed skill includes a complete here-document
130
- example for orchestrating agents.
126
+ The runner receives no conversation history and cannot be steered mid-run. Make
127
+ the handoff self-contained: state the goal, hard constraints, relevant findings
128
+ and evidence, approaches ruled out and why, and the definition of done and
129
+ verification. The installed skill includes a complete here-document example for
130
+ orchestrating agents.
131
131
 
132
132
  With no description there is nothing for a task runner to do, so only the
133
133
  reviews run — against work that already exists.
@@ -136,25 +136,32 @@ reviews run — against work that already exists.
136
136
  # Review uncommitted work before opening a pull request
137
137
  lemming exec --review readability
138
138
 
139
- # Review someone else's branch, checked out in a worktree
140
- lemming -C ../review-worktree exec --review testing --scope main...HEAD
139
+ # Review specific paths, or globs the shell should leave alone
140
+ lemming exec --review readability --scope src/api --scope 'src/**/*.ts'
141
+
142
+ # Review what a branch changed, with paths on stdin
143
+ git diff -z --name-only --relative --diff-filter=d main... |
144
+ lemming exec --review testing --scope -
141
145
 
142
146
  # Do the work, then gate it
143
147
  lemming exec "Add pagination to the tasks API" --review '*'
144
148
  ```
145
149
 
146
- `--scope` takes paths, which pass through untouched, or a git revision range,
147
- which is resolved to the files it changed. It defaults to uncommitted work, or
148
- to the whole tree outside a git repository.
150
+ `--scope` takes a path or glob, and each must match something; repeat it for
151
+ several, or pass `-` to read them from stdin. It only applies with `--review`
152
+ and is an error with a description: a task's reviews look at what the task
153
+ changed. Without `--scope`, reviews cover uncommitted work, or the whole tree
154
+ outside a git repository; a clean tree reviews nothing. Arguments for the agent
155
+ CLI go after `--`.
149
156
 
150
157
  Reviews may edit the working directory. The built-in readability review
151
- intentionally applies fixes, and testing may repair production code or tests.
152
- UX is instructed to remain advisory, but its prompt and `--no-yolo` do not
153
- enforce filesystem isolation. A global review override may perform any actions
154
- its prompt requests. There is no read-only review mode: runners keep write
155
- access, so a prompt-only flag could not be enforced. When isolation is required,
156
- follow the working-copy guidance below and pass a separate VCS-managed working
157
- copy with `lemming -C <working-copy> exec --review ...`.
158
+ intentionally applies fixes, and testing may repair production code or tests. UX
159
+ is instructed to remain advisory, but its prompt and `--no-yolo` do not enforce
160
+ filesystem isolation. A global review override may perform any actions its
161
+ prompt requests. There is no read-only review mode: runners keep write access,
162
+ so a prompt-only flag could not be enforced. When isolation is required, follow
163
+ the working-copy guidance below and pass a separate VCS-managed working copy
164
+ with `lemming -C <working-copy> exec --review ...`.
158
165
 
159
166
  Each run is self-contained: nothing is read from the project's roadmap or its
160
167
  local hooks, it runs only the requested task and reviews, and its state
@@ -164,9 +171,8 @@ printed, so the log can be read. Kept directories live in
164
171
  recent failure is always still there to inspect. Stdout carries the agent's
165
172
  message alone and everything else goes to stderr, so the output can be consumed
166
173
  directly. Stderr reports the ephemeral tasks file and task ID at startup; pass
167
- those to
168
- `lemming --tasks-file <path> status <id>` or `logs <id>` to inspect an active
169
- run. Use the global verbose flag (`lemming -v exec ...`) to stream runner
174
+ those to `lemming --tasks-file <path> status <id>` or `logs <id>` to inspect an
175
+ active run. Use the global verbose flag (`lemming -v exec ...`) to stream runner
170
176
  activity.
171
177
 
172
178
  Note that the agent runs unattended (`--yolo` by default), so it does not
@@ -206,8 +212,7 @@ lemming skill status
206
212
 
207
213
  `lemming skill uninstall` reverses the default installation sweep. Both refuse
208
214
  to touch a directory that does not hold this skill, so a mistyped `--to` fails
209
- instead of deleting work. Use `--link` on a durable installation to track
210
- package upgrades without reinstalling.
215
+ instead of deleting work.
211
216
 
212
217
  ---
213
218
 
@@ -250,14 +255,16 @@ loops through each pending task:
250
255
  with `lemming add`, breaking down complex work into smaller steps that
251
256
  Lemming will pick up automatically.
252
257
  4. **Retry or advance**: On failure, timeout, or cancellation, Lemming saves
253
- the discarded workspace diff as an `attempt-<N>.patch` artifact, restores
254
- the working tree to the task's pre-run snapshot, and retries up to the
255
- configured limit with accumulated progress as context. On success, it moves
256
- to the next task.
258
+ the discarded workspace diff as an `attempt-<N>.patch` artifact (a repeated
259
+ number gets `-2`, `-3`, ... rather than overwriting), restores the working
260
+ tree to the task's pre-run snapshot, and retries up to the configured limit
261
+ with accumulated progress as context. A timeout also keeps the tail of the
262
+ agent's output as `attempt-<N>-timeout.log`, so the retry starts from what
263
+ the killed attempt found. On success, it moves to the next task.
257
264
  5. **Orchestration**: After each task, Lemming can run one or more
258
265
  **Orchestrator Hooks** (like the built-in `roadmap` hook) to evaluate the
259
- results and adapt the roadmap if needed. Hooks are enabled by default but
260
- can be disabled via configuration.
266
+ results and adapt the roadmap if needed. Only `roadmap` runs by default;
267
+ reviews are opt-in.
261
268
 
262
269
  ---
263
270
 
@@ -385,35 +392,41 @@ These come before the subcommand and apply to all of them.
385
392
  - `--json`: Emit machine-readable JSON instead of formatted text, so scripts
386
393
  never have to parse the internal state file.
387
394
  - `--brief`: Omit task descriptions, which otherwise dominate the output.
395
+ - A task is `pending`, then `running`; once its agent finishes it is
396
+ `completing` or `failing` while hooks review it, and it ends `completed`,
397
+ `failed`, `cancelled`, or `superseded`.
398
+ - An active task with no loop running is flagged as abandoned;
399
+ `lemming run --detach` reclaims it. If its runner outlived the loop,
400
+ `status` names the runner's pid to kill first, and `run` refuses to start a
401
+ second runner until it is gone.
402
+ - A running task that has recorded no progress for a quarter of the time limit
403
+ is flagged as stale.
388
404
  - **`goal [<text>]`**: Set or view the long-term goal shared by all tasks.
389
405
  Supports `-f/--file`.
390
406
  - **`add <desc>`**: Append a new task. Supports `--index`, `--runner`,
391
407
  `--model`, `--oneshot`, and `--hooks` (the task's reviews, comma-separated;
392
408
  inherited from `--parent` when omitted).
393
- - **`edit <id>`**: Modify a task's description, runner, model, position,
394
- oneshot flag (`--oneshot/--no-oneshot`), or reviews (`--hooks`;
395
- `--project-hooks` restores the project's).
409
+ - **`edit <id>`**: Modify a task's description, runner, model, position, oneshot
410
+ flag (`--oneshot/--no-oneshot`), or reviews (`--hooks`; `--project-hooks`
411
+ restores the project's).
396
412
  - **`brief <id> [text]`**: View or set a task's long-form brief. Unlike the
397
413
  description it has no length cap, and it is appended to the runner prompt
398
414
  automatically — the right home for measured timings, exact failing selectors,
399
415
  or why a previous attempt was wrong. Supports `-f/--file`.
400
- - **`artifact <id> [name] [text]`**: Store or read a task's verbose
401
- diagnostics — stack traces, test dumps, long logs — outside `tasks.yml`, so
402
- they never bloat the prompts of later attempts. With no name it lists what the
403
- task has stored; with `--note` it also records the one-line progress pointer
404
- for you. Later attempts see the artifact names in their prompt automatically.
405
- Supports `-f/--file`.
416
+ - **`artifact <id> [name] [text]`**: Store or read a task's verbose diagnostics
417
+ — stack traces, test dumps, long logs — outside `tasks.yml`, so they never
418
+ bloat the prompts of later attempts. With no name it lists what the task has
419
+ stored; with `--note` it also records the one-line progress pointer for you.
420
+ Later attempts see the artifact names in their prompt automatically. Supports
421
+ `-f/--file`.
406
422
  - **`delete <id>`**: Remove an unstarted task while retaining its runner log;
407
423
  its artifacts go with it. Tasks with execution history require `--force`;
408
424
  autonomous restructuring should use `supersede`. Supports `--all` and
409
425
  `--completed` for bulk cleanup, including logs.
410
426
  - **`supersede <id> --reason <text>`**: Retire a replaced or split task without
411
427
  losing its progress, timings, log, or links to replacement tasks.
412
- - **`progress`**: Manage progress entries and findings for specific tasks.
413
- - `list <id>`: List all progress for a task.
414
- - `add <id> <finding>`: Record a new technical detail.
415
- - `edit <id> <index> <text>`: Modify an existing progress entry.
416
- - `delete <id> <index>`: Remove a progress entry.
428
+ - **`progress <id> [text]`**: Record a progress entry or finding for a task;
429
+ `status <id>` lists them. Supports `-f/--file`.
417
430
  - **`config`**: Manage project configuration (runner, model, retries, time
418
431
  limit).
419
432
  - `list`: View current configuration.
@@ -427,20 +440,22 @@ These come before the subcommand and apply to all of them.
427
440
  [readability](https://github.com/owahltinez/readability) package. Ruff and
428
441
  Pyrefly run with bundled Google-style defaults unless the target project
429
442
  defines its own configuration.
430
- - `check <paths>...`: Run formatters and linters.
443
+ - `check <paths>...`: Run formatters and linters; exits non-zero on any
444
+ finding. `--fix` applies safe fixes, `--unsafe` also behavior-changing ones.
431
445
  - `guide <language>`: Fetch and view style guides.
432
- - `languages`: List all supported languages.
433
- - `sync`: Synchronize style guides from the web.
446
+ - `sync`: Refetch style guides from the web.
434
447
 
435
448
  ### Task Status
436
449
 
437
- - **`complete <id>`**: Mark a task as successful.
450
+ - **`complete <id>`**: Mark a task as successful; requires recorded progress.
451
+ `--force` finalizes an in-progress task at once, without running hooks.
452
+ - **`uncomplete <id>`**: Move a completed task back to pending.
438
453
  - **`fail <id>`**: Mark a task as a terminal failure (will not be retried).
439
454
  - **`reject <id> <reason>`**: Reject a finalizing task's completion from a
440
455
  review hook, sending it back for a retry.
441
- - **`cancel <id>`**: Stop an in-progress task (kills the runner process and
442
- restores its pre-run workspace snapshot).
443
- - **`reset <id>`**: Clear attempts and progress to start a task fresh.
456
+ - **`cancel <id>`**: Cancel a pending or in-progress task (an active one's
457
+ runner is killed and its pre-run workspace snapshot restored).
458
+ - **`reset <id>`**: Clear attempts, progress, and logs to start a task fresh.
444
459
  - **`diff [<id>]`**: Print the workspace diff captured for a task (defaults to
445
460
  the active or most recent snapshotted task).
446
461
  - `--stat`: Show diffstat summary instead of the full patch.
@@ -467,15 +482,20 @@ These come before the subcommand and apply to all of them.
467
482
  - `--env`: Set environment variables for the runner (e.g., `--env KEY=VALUE`).
468
483
  - `--no-defaults`: Skip default flag injection for known runners.
469
484
  - `--until <task-id>`: Stop between tasks once that task has completed or
470
- failed, printing `Reached milestone <id>; M pending.` so a supervising
471
- agent can review before queueing the next milestone. Tasks inserted ahead
472
- of it still run; if the roadmap hook supersedes it, the run continues until
473
- every replacement has settled. Deleting it stops the run with exit 1.
485
+ failed, printing `Reached milestone <id>; M pending.` so a supervising agent
486
+ can review before queueing the next milestone. Tasks inserted ahead of it
487
+ still run; if the roadmap hook supersedes it, the run continues until every
488
+ replacement has settled. Deleting it stops the run with exit 1.
474
489
  - `--max-tasks N`: Stop between tasks once `N` have completed or failed,
475
490
  printing `Stopped after N tasks; M pending.` Attempts that leave a task
476
491
  pending do not count. Combine with `--until` as a budget cap.
477
- - `--`: Use `--` to pass any flag directly to the underlying runner. A
478
- per-task `--runner`/`--model` overrides anything passed here.
492
+ - `--detach`: Start the loop in its own session and return, printing its PID
493
+ and log file (`loop.log` in the project's state directory; one over 1 MB
494
+ moves to `loop.log.1` first). The loop outlives the shell or agent harness
495
+ that started it; check on it with `status --brief` and end it with `stop`.
496
+ - `--`: Arguments after `--` go directly to the underlying runner; any other
497
+ extra argument is an error. A per-task `--runner`/`--model` overrides
498
+ anything passed here.
479
499
  - **`exec [<description>]`**: Run one task, or one set of reviews, outside any
480
500
  roadmap. Prints the agent's closing message to stdout and everything else to
481
501
  stderr, including its tasks file and task ID at startup. Use
@@ -487,9 +507,11 @@ These come before the subcommand and apply to all of them.
487
507
  repeated; `'*'` selects every one. With no description, only the reviews
488
508
  run. Reviews may edit the working directory. Hooks that revise the roadmap
489
509
  cannot be selected.
490
- - `--scope <path|range>`: What the reviews look at. Paths pass through; a git
491
- revision range is resolved to the files it changed. Defaults to uncommitted
492
- work, or the whole tree outside a repository.
510
+ - `--scope <path>`: With `--review` and no description, a path or glob to
511
+ review instead of uncommitted work; an error with a description. Each must
512
+ match something. Repeat it for several, or pass `-` to read them from stdin:
513
+ NUL-separated if it holds any NUL, as from `git diff -z`, or else one per
514
+ line.
493
515
  - `--runner`, `--model`: Which agent CLI and model to use.
494
516
  - `--retries`: Maximum total task-runner attempts, including the first
495
517
  (default 1). Does not retry review-only runs.
@@ -497,11 +519,12 @@ These come before the subcommand and apply to all of them.
497
519
  limit).
498
520
  - `--yolo/--no-yolo`: Run the agent unattended (default: True).
499
521
  - `--keep`: Keep the run's state directory even when it succeeds.
522
+ - `--`: Everything after it goes to the agent CLI; any other extra argument is
523
+ an error.
500
524
  - **`skill install`**: Install the packaged Agent Skill so agents discover
501
525
  Lemming. Writes to `~/.agents/skills` and every detected tool-specific
502
526
  directory by default. Re-running it refreshes existing Lemming copies.
503
527
  - `--to <dir>`: Install into a specific skills directory.
504
- - `--link`: Symlink instead of copying, so upgrades take effect immediately.
505
528
  - `--dry-run`: Print what would happen, refusals included.
506
529
  - **`skill uninstall`**: Remove it again. Supports `--to` and `--dry-run`.
507
530
  - **`skill status`**: Show every supported location and installation state.
@@ -75,7 +75,7 @@ Start the autonomous loop.
75
75
  # claude, or codex)
76
76
  lemming run
77
77
 
78
- # Flags passed after -- are sent directly to the underlying runner
78
+ # Arguments after -- are sent directly to the underlying runner
79
79
  lemming run -- --model claude-3-5-sonnet
80
80
  ```
81
81
 
@@ -103,11 +103,11 @@ lemming exec "Fix the flaky test in runner_test.py" --retries 3
103
103
  cat handoff.md | lemming -v exec -f - --runner agy
104
104
  ```
105
105
 
106
- The runner receives no conversation history and cannot be steered mid-run.
107
- Make the handoff self-contained: state the goal, hard constraints, relevant
108
- findings and evidence, approaches ruled out and why, and the definition of done
109
- and verification. The installed skill includes a complete here-document
110
- example for orchestrating agents.
106
+ The runner receives no conversation history and cannot be steered mid-run. Make
107
+ the handoff self-contained: state the goal, hard constraints, relevant findings
108
+ and evidence, approaches ruled out and why, and the definition of done and
109
+ verification. The installed skill includes a complete here-document example for
110
+ orchestrating agents.
111
111
 
112
112
  With no description there is nothing for a task runner to do, so only the
113
113
  reviews run — against work that already exists.
@@ -116,25 +116,32 @@ reviews run — against work that already exists.
116
116
  # Review uncommitted work before opening a pull request
117
117
  lemming exec --review readability
118
118
 
119
- # Review someone else's branch, checked out in a worktree
120
- lemming -C ../review-worktree exec --review testing --scope main...HEAD
119
+ # Review specific paths, or globs the shell should leave alone
120
+ lemming exec --review readability --scope src/api --scope 'src/**/*.ts'
121
+
122
+ # Review what a branch changed, with paths on stdin
123
+ git diff -z --name-only --relative --diff-filter=d main... |
124
+ lemming exec --review testing --scope -
121
125
 
122
126
  # Do the work, then gate it
123
127
  lemming exec "Add pagination to the tasks API" --review '*'
124
128
  ```
125
129
 
126
- `--scope` takes paths, which pass through untouched, or a git revision range,
127
- which is resolved to the files it changed. It defaults to uncommitted work, or
128
- to the whole tree outside a git repository.
130
+ `--scope` takes a path or glob, and each must match something; repeat it for
131
+ several, or pass `-` to read them from stdin. It only applies with `--review`
132
+ and is an error with a description: a task's reviews look at what the task
133
+ changed. Without `--scope`, reviews cover uncommitted work, or the whole tree
134
+ outside a git repository; a clean tree reviews nothing. Arguments for the agent
135
+ CLI go after `--`.
129
136
 
130
137
  Reviews may edit the working directory. The built-in readability review
131
- intentionally applies fixes, and testing may repair production code or tests.
132
- UX is instructed to remain advisory, but its prompt and `--no-yolo` do not
133
- enforce filesystem isolation. A global review override may perform any actions
134
- its prompt requests. There is no read-only review mode: runners keep write
135
- access, so a prompt-only flag could not be enforced. When isolation is required,
136
- follow the working-copy guidance below and pass a separate VCS-managed working
137
- copy with `lemming -C <working-copy> exec --review ...`.
138
+ intentionally applies fixes, and testing may repair production code or tests. UX
139
+ is instructed to remain advisory, but its prompt and `--no-yolo` do not enforce
140
+ filesystem isolation. A global review override may perform any actions its
141
+ prompt requests. There is no read-only review mode: runners keep write access,
142
+ so a prompt-only flag could not be enforced. When isolation is required, follow
143
+ the working-copy guidance below and pass a separate VCS-managed working copy
144
+ with `lemming -C <working-copy> exec --review ...`.
138
145
 
139
146
  Each run is self-contained: nothing is read from the project's roadmap or its
140
147
  local hooks, it runs only the requested task and reviews, and its state
@@ -144,9 +151,8 @@ printed, so the log can be read. Kept directories live in
144
151
  recent failure is always still there to inspect. Stdout carries the agent's
145
152
  message alone and everything else goes to stderr, so the output can be consumed
146
153
  directly. Stderr reports the ephemeral tasks file and task ID at startup; pass
147
- those to
148
- `lemming --tasks-file <path> status <id>` or `logs <id>` to inspect an active
149
- run. Use the global verbose flag (`lemming -v exec ...`) to stream runner
154
+ those to `lemming --tasks-file <path> status <id>` or `logs <id>` to inspect an
155
+ active run. Use the global verbose flag (`lemming -v exec ...`) to stream runner
150
156
  activity.
151
157
 
152
158
  Note that the agent runs unattended (`--yolo` by default), so it does not
@@ -186,8 +192,7 @@ lemming skill status
186
192
 
187
193
  `lemming skill uninstall` reverses the default installation sweep. Both refuse
188
194
  to touch a directory that does not hold this skill, so a mistyped `--to` fails
189
- instead of deleting work. Use `--link` on a durable installation to track
190
- package upgrades without reinstalling.
195
+ instead of deleting work.
191
196
 
192
197
  ---
193
198
 
@@ -230,14 +235,16 @@ loops through each pending task:
230
235
  with `lemming add`, breaking down complex work into smaller steps that
231
236
  Lemming will pick up automatically.
232
237
  4. **Retry or advance**: On failure, timeout, or cancellation, Lemming saves
233
- the discarded workspace diff as an `attempt-<N>.patch` artifact, restores
234
- the working tree to the task's pre-run snapshot, and retries up to the
235
- configured limit with accumulated progress as context. On success, it moves
236
- to the next task.
238
+ the discarded workspace diff as an `attempt-<N>.patch` artifact (a repeated
239
+ number gets `-2`, `-3`, ... rather than overwriting), restores the working
240
+ tree to the task's pre-run snapshot, and retries up to the configured limit
241
+ with accumulated progress as context. A timeout also keeps the tail of the
242
+ agent's output as `attempt-<N>-timeout.log`, so the retry starts from what
243
+ the killed attempt found. On success, it moves to the next task.
237
244
  5. **Orchestration**: After each task, Lemming can run one or more
238
245
  **Orchestrator Hooks** (like the built-in `roadmap` hook) to evaluate the
239
- results and adapt the roadmap if needed. Hooks are enabled by default but
240
- can be disabled via configuration.
246
+ results and adapt the roadmap if needed. Only `roadmap` runs by default;
247
+ reviews are opt-in.
241
248
 
242
249
  ---
243
250
 
@@ -365,35 +372,41 @@ These come before the subcommand and apply to all of them.
365
372
  - `--json`: Emit machine-readable JSON instead of formatted text, so scripts
366
373
  never have to parse the internal state file.
367
374
  - `--brief`: Omit task descriptions, which otherwise dominate the output.
375
+ - A task is `pending`, then `running`; once its agent finishes it is
376
+ `completing` or `failing` while hooks review it, and it ends `completed`,
377
+ `failed`, `cancelled`, or `superseded`.
378
+ - An active task with no loop running is flagged as abandoned;
379
+ `lemming run --detach` reclaims it. If its runner outlived the loop,
380
+ `status` names the runner's pid to kill first, and `run` refuses to start a
381
+ second runner until it is gone.
382
+ - A running task that has recorded no progress for a quarter of the time limit
383
+ is flagged as stale.
368
384
  - **`goal [<text>]`**: Set or view the long-term goal shared by all tasks.
369
385
  Supports `-f/--file`.
370
386
  - **`add <desc>`**: Append a new task. Supports `--index`, `--runner`,
371
387
  `--model`, `--oneshot`, and `--hooks` (the task's reviews, comma-separated;
372
388
  inherited from `--parent` when omitted).
373
- - **`edit <id>`**: Modify a task's description, runner, model, position,
374
- oneshot flag (`--oneshot/--no-oneshot`), or reviews (`--hooks`;
375
- `--project-hooks` restores the project's).
389
+ - **`edit <id>`**: Modify a task's description, runner, model, position, oneshot
390
+ flag (`--oneshot/--no-oneshot`), or reviews (`--hooks`; `--project-hooks`
391
+ restores the project's).
376
392
  - **`brief <id> [text]`**: View or set a task's long-form brief. Unlike the
377
393
  description it has no length cap, and it is appended to the runner prompt
378
394
  automatically — the right home for measured timings, exact failing selectors,
379
395
  or why a previous attempt was wrong. Supports `-f/--file`.
380
- - **`artifact <id> [name] [text]`**: Store or read a task's verbose
381
- diagnostics — stack traces, test dumps, long logs — outside `tasks.yml`, so
382
- they never bloat the prompts of later attempts. With no name it lists what the
383
- task has stored; with `--note` it also records the one-line progress pointer
384
- for you. Later attempts see the artifact names in their prompt automatically.
385
- Supports `-f/--file`.
396
+ - **`artifact <id> [name] [text]`**: Store or read a task's verbose diagnostics
397
+ — stack traces, test dumps, long logs — outside `tasks.yml`, so they never
398
+ bloat the prompts of later attempts. With no name it lists what the task has
399
+ stored; with `--note` it also records the one-line progress pointer for you.
400
+ Later attempts see the artifact names in their prompt automatically. Supports
401
+ `-f/--file`.
386
402
  - **`delete <id>`**: Remove an unstarted task while retaining its runner log;
387
403
  its artifacts go with it. Tasks with execution history require `--force`;
388
404
  autonomous restructuring should use `supersede`. Supports `--all` and
389
405
  `--completed` for bulk cleanup, including logs.
390
406
  - **`supersede <id> --reason <text>`**: Retire a replaced or split task without
391
407
  losing its progress, timings, log, or links to replacement tasks.
392
- - **`progress`**: Manage progress entries and findings for specific tasks.
393
- - `list <id>`: List all progress for a task.
394
- - `add <id> <finding>`: Record a new technical detail.
395
- - `edit <id> <index> <text>`: Modify an existing progress entry.
396
- - `delete <id> <index>`: Remove a progress entry.
408
+ - **`progress <id> [text]`**: Record a progress entry or finding for a task;
409
+ `status <id>` lists them. Supports `-f/--file`.
397
410
  - **`config`**: Manage project configuration (runner, model, retries, time
398
411
  limit).
399
412
  - `list`: View current configuration.
@@ -407,20 +420,22 @@ These come before the subcommand and apply to all of them.
407
420
  [readability](https://github.com/owahltinez/readability) package. Ruff and
408
421
  Pyrefly run with bundled Google-style defaults unless the target project
409
422
  defines its own configuration.
410
- - `check <paths>...`: Run formatters and linters.
423
+ - `check <paths>...`: Run formatters and linters; exits non-zero on any
424
+ finding. `--fix` applies safe fixes, `--unsafe` also behavior-changing ones.
411
425
  - `guide <language>`: Fetch and view style guides.
412
- - `languages`: List all supported languages.
413
- - `sync`: Synchronize style guides from the web.
426
+ - `sync`: Refetch style guides from the web.
414
427
 
415
428
  ### Task Status
416
429
 
417
- - **`complete <id>`**: Mark a task as successful.
430
+ - **`complete <id>`**: Mark a task as successful; requires recorded progress.
431
+ `--force` finalizes an in-progress task at once, without running hooks.
432
+ - **`uncomplete <id>`**: Move a completed task back to pending.
418
433
  - **`fail <id>`**: Mark a task as a terminal failure (will not be retried).
419
434
  - **`reject <id> <reason>`**: Reject a finalizing task's completion from a
420
435
  review hook, sending it back for a retry.
421
- - **`cancel <id>`**: Stop an in-progress task (kills the runner process and
422
- restores its pre-run workspace snapshot).
423
- - **`reset <id>`**: Clear attempts and progress to start a task fresh.
436
+ - **`cancel <id>`**: Cancel a pending or in-progress task (an active one's
437
+ runner is killed and its pre-run workspace snapshot restored).
438
+ - **`reset <id>`**: Clear attempts, progress, and logs to start a task fresh.
424
439
  - **`diff [<id>]`**: Print the workspace diff captured for a task (defaults to
425
440
  the active or most recent snapshotted task).
426
441
  - `--stat`: Show diffstat summary instead of the full patch.
@@ -447,15 +462,20 @@ These come before the subcommand and apply to all of them.
447
462
  - `--env`: Set environment variables for the runner (e.g., `--env KEY=VALUE`).
448
463
  - `--no-defaults`: Skip default flag injection for known runners.
449
464
  - `--until <task-id>`: Stop between tasks once that task has completed or
450
- failed, printing `Reached milestone <id>; M pending.` so a supervising
451
- agent can review before queueing the next milestone. Tasks inserted ahead
452
- of it still run; if the roadmap hook supersedes it, the run continues until
453
- every replacement has settled. Deleting it stops the run with exit 1.
465
+ failed, printing `Reached milestone <id>; M pending.` so a supervising agent
466
+ can review before queueing the next milestone. Tasks inserted ahead of it
467
+ still run; if the roadmap hook supersedes it, the run continues until every
468
+ replacement has settled. Deleting it stops the run with exit 1.
454
469
  - `--max-tasks N`: Stop between tasks once `N` have completed or failed,
455
470
  printing `Stopped after N tasks; M pending.` Attempts that leave a task
456
471
  pending do not count. Combine with `--until` as a budget cap.
457
- - `--`: Use `--` to pass any flag directly to the underlying runner. A
458
- per-task `--runner`/`--model` overrides anything passed here.
472
+ - `--detach`: Start the loop in its own session and return, printing its PID
473
+ and log file (`loop.log` in the project's state directory; one over 1 MB
474
+ moves to `loop.log.1` first). The loop outlives the shell or agent harness
475
+ that started it; check on it with `status --brief` and end it with `stop`.
476
+ - `--`: Arguments after `--` go directly to the underlying runner; any other
477
+ extra argument is an error. A per-task `--runner`/`--model` overrides
478
+ anything passed here.
459
479
  - **`exec [<description>]`**: Run one task, or one set of reviews, outside any
460
480
  roadmap. Prints the agent's closing message to stdout and everything else to
461
481
  stderr, including its tasks file and task ID at startup. Use
@@ -467,9 +487,11 @@ These come before the subcommand and apply to all of them.
467
487
  repeated; `'*'` selects every one. With no description, only the reviews
468
488
  run. Reviews may edit the working directory. Hooks that revise the roadmap
469
489
  cannot be selected.
470
- - `--scope <path|range>`: What the reviews look at. Paths pass through; a git
471
- revision range is resolved to the files it changed. Defaults to uncommitted
472
- work, or the whole tree outside a repository.
490
+ - `--scope <path>`: With `--review` and no description, a path or glob to
491
+ review instead of uncommitted work; an error with a description. Each must
492
+ match something. Repeat it for several, or pass `-` to read them from stdin:
493
+ NUL-separated if it holds any NUL, as from `git diff -z`, or else one per
494
+ line.
473
495
  - `--runner`, `--model`: Which agent CLI and model to use.
474
496
  - `--retries`: Maximum total task-runner attempts, including the first
475
497
  (default 1). Does not retry review-only runs.
@@ -477,11 +499,12 @@ These come before the subcommand and apply to all of them.
477
499
  limit).
478
500
  - `--yolo/--no-yolo`: Run the agent unattended (default: True).
479
501
  - `--keep`: Keep the run's state directory even when it succeeds.
502
+ - `--`: Everything after it goes to the agent CLI; any other extra argument is
503
+ an error.
480
504
  - **`skill install`**: Install the packaged Agent Skill so agents discover
481
505
  Lemming. Writes to `~/.agents/skills` and every detected tool-specific
482
506
  directory by default. Re-running it refreshes existing Lemming copies.
483
507
  - `--to <dir>`: Install into a specific skills directory.
484
- - `--link`: Symlink instead of copying, so upgrades take effect immediately.
485
508
  - `--dry-run`: Print what would happen, refusals included.
486
509
  - **`skill uninstall`**: Remove it again. Supports `--to` and `--dry-run`.
487
510
  - **`skill status`**: Show every supported location and installation state.