aidd-cli 0.1.0__tar.gz → 0.5.0__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.
- aidd_cli-0.5.0/MANIFEST.in +3 -0
- {aidd_cli-0.1.0 → aidd_cli-0.5.0}/PKG-INFO +129 -17
- {aidd_cli-0.1.0 → aidd_cli-0.5.0}/README.md +128 -16
- aidd_cli-0.5.0/adapters/AGENTS.md.snippet +29 -0
- aidd_cli-0.5.0/adapters/GEMINI.md +16 -0
- aidd_cli-0.5.0/adapters/README.md +241 -0
- aidd_cli-0.5.0/adapters/antigravity/rules/aidd.md +29 -0
- aidd_cli-0.5.0/adapters/antigravity/skills/aidd/SKILL.md +30 -0
- aidd_cli-0.5.0/adapters/dot-aidd/AIDD.md +708 -0
- aidd_cli-0.5.0/adapters/dot-aidd/CORE.md +133 -0
- aidd_cli-0.5.0/adapters/dot-aidd/extensions/azure_devops/manifest.json +9 -0
- aidd_cli-0.5.0/adapters/dot-aidd/extensions/azure_devops/provider.py +121 -0
- aidd_cli-0.5.0/adapters/dot-aidd/extensions/bitbucket/manifest.json +9 -0
- aidd_cli-0.5.0/adapters/dot-aidd/extensions/bitbucket/provider.py +138 -0
- aidd_cli-0.5.0/adapters/dot-aidd/extensions/github/manifest.json +9 -0
- aidd_cli-0.5.0/adapters/dot-aidd/extensions/github/provider.py +82 -0
- aidd_cli-0.5.0/adapters/dot-aidd/hooks/_common.py +197 -0
- aidd_cli-0.5.0/adapters/dot-aidd/hooks/codex_patch.py +181 -0
- aidd_cli-0.5.0/adapters/dot-aidd/hooks/codex_subagent.py +63 -0
- aidd_cli-0.5.0/adapters/dot-aidd/hooks/host_adapter.py +696 -0
- aidd_cli-0.5.0/adapters/dot-aidd/hooks/mark_agent_dispatch.py +169 -0
- aidd_cli-0.5.0/adapters/dot-aidd/hooks/mark_code_edit.py +141 -0
- aidd_cli-0.5.0/adapters/dot-aidd/hooks/record_dispatch_pre.py +57 -0
- aidd_cli-0.5.0/adapters/dot-aidd/hooks/require_aidd.py +66 -0
- aidd_cli-0.5.0/adapters/dot-aidd/hooks/require_graph_coherence_audit.py +150 -0
- aidd_cli-0.5.0/adapters/dot-aidd/hooks/require_independent_audit.py +83 -0
- aidd_cli-0.5.0/adapters/dot-aidd/hooks/rule_gate.py +2120 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_amendment_verify.py +489 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_calibrate.py +135 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_close_ledger.py +340 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_coverage.py +573 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_doctor.py +656 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_embed.py +44 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_evidence.py +2446 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_gitignore.py +57 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_graphs.py +1284 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_guard.py +729 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_job.py +726 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_memory.py +1379 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_memory_import.py +403 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_navigator.py +196 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_next.py +230 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_pending.py +173 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_progress.py +510 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_provider_attestation.py +526 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_readiness.py +128 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_red.py +279 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_reuse.py +104 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_review.py +1756 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_review_items.py +1042 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_review_state.py +627 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_round.py +773 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_rules.py +3909 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_source_shape.py +137 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_sqlcheck.py +268 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_status.py +4081 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_sync.py +391 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_tools.py +345 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_triage.py +132 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_unit_audit.py +795 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_unit_consent.py +54 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_unit_visual.py +240 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/aidd_units.py +782 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/alias_graph.py +153 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/check_charter.py +249 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/check_spec.py +731 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/code_graph.py +1062 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/component_graph.py +536 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/extension_registry.py +309 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/find_spec.py +1315 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/flowmap.py +1542 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/graph_explorer.py +1644 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/graph_snapshot.py +1098 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/graph_trace.py +733 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/link_pr_to_task.py +125 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/method_graph.py +297 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/sql_graph.py +1125 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/sync_issues.py +117 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/tasks_to_issues.py +283 -0
- aidd_cli-0.5.0/adapters/dot-aidd/scripts/unified_graph.py +405 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/STATE.md +41 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/calibration.toon +6 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/charter.md +96 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/comprehensive-documentation.md +26 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/contracts.md +31 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/data-model.md +58 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/design-system/MASTER.md +31 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/design-system/components-index.md +13 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/design-system/page-override.md +8 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/gitignore.aidd +9 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/mockup-audit.md +60 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/navigator.html +32 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/perf/README.md +22 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/perf/plan-risk.test.sql +72 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/perf/seed-generate-series.sql +52 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/plan.md +62 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/qa-audit.md +83 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/research.md +8 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/review-full.html +195 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/review.html +1288 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/spec.md +162 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/tasks.md +102 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/traceability.md +21 -0
- aidd_cli-0.5.0/adapters/dot-aidd/templates/visual-flow.toon +28 -0
- aidd_cli-0.5.0/adapters/opencode/plugins/aidd.js +116 -0
- aidd_cli-0.5.0/adapters/opencode/skills/aidd/SKILL.md +33 -0
- {aidd_cli-0.1.0 → aidd_cli-0.5.0}/aidd/__init__.py +1 -1
- aidd_cli-0.5.0/aidd/cli.py +730 -0
- {aidd_cli-0.1.0 → aidd_cli-0.5.0}/aidd_cli.egg-info/PKG-INFO +129 -17
- aidd_cli-0.5.0/aidd_cli.egg-info/SOURCES.txt +236 -0
- {aidd_cli-0.1.0 → aidd_cli-0.5.0}/pyproject.toml +1 -1
- aidd_cli-0.5.0/setup.py +32 -0
- aidd_cli-0.5.0/skill/AIDD.md +708 -0
- aidd_cli-0.5.0/skill/SKILL.md +553 -0
- aidd_cli-0.5.0/skill/extensions/adapters/cline/manifest.json +14 -0
- aidd_cli-0.5.0/skill/extensions/adapters/copilot/manifest.json +15 -0
- aidd_cli-0.5.0/skill/extensions/adapters/cursor/manifest.json +14 -0
- aidd_cli-0.5.0/skill/extensions/adapters/gemini/manifest.json +14 -0
- aidd_cli-0.5.0/skill/extensions/adapters/windsurf/manifest.json +14 -0
- aidd_cli-0.5.0/skill/extensions/azure_devops/manifest.json +9 -0
- aidd_cli-0.5.0/skill/extensions/azure_devops/provider.py +121 -0
- aidd_cli-0.5.0/skill/extensions/bitbucket/manifest.json +9 -0
- aidd_cli-0.5.0/skill/extensions/bitbucket/provider.py +138 -0
- aidd_cli-0.5.0/skill/extensions/github/manifest.json +9 -0
- aidd_cli-0.5.0/skill/extensions/github/provider.py +82 -0
- aidd_cli-0.5.0/skill/hooks/_common.py +197 -0
- aidd_cli-0.5.0/skill/hooks/codex_patch.py +181 -0
- aidd_cli-0.5.0/skill/hooks/codex_subagent.py +63 -0
- aidd_cli-0.5.0/skill/hooks/graph_refresh.py +98 -0
- aidd_cli-0.5.0/skill/hooks/host_adapter.py +696 -0
- aidd_cli-0.5.0/skill/hooks/mark_agent_dispatch.py +169 -0
- aidd_cli-0.5.0/skill/hooks/mark_code_edit.py +141 -0
- aidd_cli-0.5.0/skill/hooks/mark_graph_rebuild.py +76 -0
- aidd_cli-0.5.0/skill/hooks/mark_invoked.py +31 -0
- aidd_cli-0.5.0/skill/hooks/mark_user_question.py +82 -0
- aidd_cli-0.5.0/skill/hooks/memory_context.py +48 -0
- aidd_cli-0.5.0/skill/hooks/memory_file_context.py +92 -0
- aidd_cli-0.5.0/skill/hooks/prompt_trigger.py +330 -0
- aidd_cli-0.5.0/skill/hooks/read_hint.py +295 -0
- aidd_cli-0.5.0/skill/hooks/record_dispatch_pre.py +57 -0
- aidd_cli-0.5.0/skill/hooks/require_aidd.py +66 -0
- aidd_cli-0.5.0/skill/hooks/require_graph_coherence_audit.py +150 -0
- aidd_cli-0.5.0/skill/hooks/require_independent_audit.py +83 -0
- aidd_cli-0.5.0/skill/hooks/rule_gate.py +2120 -0
- aidd_cli-0.5.0/skill/hooks/session_start.py +67 -0
- aidd_cli-0.5.0/skill/hooks/stop_gate.py +274 -0
- aidd_cli-0.5.0/skill/hooks/tool_first_hint.py +284 -0
- aidd_cli-0.5.0/skill/scripts/aidd_amendment_verify.py +489 -0
- aidd_cli-0.5.0/skill/scripts/aidd_calibrate.py +135 -0
- aidd_cli-0.5.0/skill/scripts/aidd_close_ledger.py +340 -0
- aidd_cli-0.5.0/skill/scripts/aidd_coverage.py +573 -0
- aidd_cli-0.5.0/skill/scripts/aidd_doctor.py +656 -0
- aidd_cli-0.5.0/skill/scripts/aidd_embed.py +44 -0
- aidd_cli-0.5.0/skill/scripts/aidd_evidence.py +2446 -0
- aidd_cli-0.5.0/skill/scripts/aidd_gitignore.py +57 -0
- aidd_cli-0.5.0/skill/scripts/aidd_graphs.py +1284 -0
- aidd_cli-0.5.0/skill/scripts/aidd_guard.py +729 -0
- aidd_cli-0.5.0/skill/scripts/aidd_job.py +726 -0
- aidd_cli-0.5.0/skill/scripts/aidd_memory.py +1379 -0
- aidd_cli-0.5.0/skill/scripts/aidd_memory_import.py +403 -0
- aidd_cli-0.5.0/skill/scripts/aidd_navigator.py +196 -0
- aidd_cli-0.5.0/skill/scripts/aidd_next.py +230 -0
- aidd_cli-0.5.0/skill/scripts/aidd_pending.py +173 -0
- aidd_cli-0.5.0/skill/scripts/aidd_progress.py +510 -0
- aidd_cli-0.5.0/skill/scripts/aidd_provider_attestation.py +526 -0
- aidd_cli-0.5.0/skill/scripts/aidd_readiness.py +128 -0
- aidd_cli-0.5.0/skill/scripts/aidd_red.py +279 -0
- aidd_cli-0.5.0/skill/scripts/aidd_reuse.py +104 -0
- aidd_cli-0.5.0/skill/scripts/aidd_review.py +1756 -0
- aidd_cli-0.5.0/skill/scripts/aidd_review_items.py +1042 -0
- aidd_cli-0.5.0/skill/scripts/aidd_review_state.py +627 -0
- aidd_cli-0.5.0/skill/scripts/aidd_round.py +773 -0
- aidd_cli-0.5.0/skill/scripts/aidd_rules.py +3909 -0
- aidd_cli-0.5.0/skill/scripts/aidd_source_shape.py +137 -0
- aidd_cli-0.5.0/skill/scripts/aidd_sqlcheck.py +268 -0
- aidd_cli-0.5.0/skill/scripts/aidd_status.py +4081 -0
- aidd_cli-0.5.0/skill/scripts/aidd_sync.py +391 -0
- aidd_cli-0.5.0/skill/scripts/aidd_tools.py +345 -0
- aidd_cli-0.5.0/skill/scripts/aidd_triage.py +132 -0
- aidd_cli-0.5.0/skill/scripts/aidd_unit_audit.py +795 -0
- aidd_cli-0.5.0/skill/scripts/aidd_unit_consent.py +54 -0
- aidd_cli-0.5.0/skill/scripts/aidd_unit_visual.py +240 -0
- aidd_cli-0.5.0/skill/scripts/aidd_units.py +782 -0
- aidd_cli-0.5.0/skill/scripts/alias_graph.py +153 -0
- aidd_cli-0.5.0/skill/scripts/check_charter.py +249 -0
- aidd_cli-0.5.0/skill/scripts/check_spec.py +731 -0
- aidd_cli-0.5.0/skill/scripts/code_graph.py +1062 -0
- aidd_cli-0.5.0/skill/scripts/component_graph.py +536 -0
- aidd_cli-0.5.0/skill/scripts/extension_registry.py +287 -0
- aidd_cli-0.5.0/skill/scripts/find_spec.py +1315 -0
- aidd_cli-0.5.0/skill/scripts/flowmap.py +1542 -0
- aidd_cli-0.5.0/skill/scripts/generate_adapters.py +221 -0
- aidd_cli-0.5.0/skill/scripts/graph_explorer.py +1644 -0
- aidd_cli-0.5.0/skill/scripts/graph_snapshot.py +1098 -0
- aidd_cli-0.5.0/skill/scripts/graph_trace.py +733 -0
- aidd_cli-0.5.0/skill/scripts/install_ci.py +114 -0
- aidd_cli-0.5.0/skill/scripts/install_hooks.py +263 -0
- aidd_cli-0.5.0/skill/scripts/install_host.py +740 -0
- aidd_cli-0.5.0/skill/scripts/link_pr_to_task.py +125 -0
- aidd_cli-0.5.0/skill/scripts/marketplace.py +262 -0
- aidd_cli-0.5.0/skill/scripts/method_graph.py +297 -0
- aidd_cli-0.5.0/skill/scripts/research_project.py +232 -0
- aidd_cli-0.5.0/skill/scripts/sql_graph.py +1125 -0
- aidd_cli-0.5.0/skill/scripts/sync_issues.py +117 -0
- aidd_cli-0.5.0/skill/scripts/tasks_to_issues.py +283 -0
- aidd_cli-0.5.0/skill/scripts/unified_graph.py +405 -0
- aidd_cli-0.5.0/skill/scripts/validate_catalog_entry.py +224 -0
- aidd_cli-0.5.0/skill/templates/STATE.md +41 -0
- aidd_cli-0.5.0/skill/templates/calibration.toon +6 -0
- aidd_cli-0.5.0/skill/templates/charter.md +96 -0
- aidd_cli-0.5.0/skill/templates/ci/azure-pipelines-aidd.yml +219 -0
- aidd_cli-0.5.0/skill/templates/ci/github-actions-aidd.yml +183 -0
- aidd_cli-0.5.0/skill/templates/comprehensive-documentation.md +26 -0
- aidd_cli-0.5.0/skill/templates/contracts.md +31 -0
- aidd_cli-0.5.0/skill/templates/data-model.md +58 -0
- aidd_cli-0.5.0/skill/templates/design-system/MASTER.md +31 -0
- aidd_cli-0.5.0/skill/templates/design-system/components-index.md +13 -0
- aidd_cli-0.5.0/skill/templates/design-system/page-override.md +8 -0
- aidd_cli-0.5.0/skill/templates/gitignore.aidd +9 -0
- aidd_cli-0.5.0/skill/templates/mockup-audit.md +60 -0
- aidd_cli-0.5.0/skill/templates/navigator.html +32 -0
- aidd_cli-0.5.0/skill/templates/perf/README.md +22 -0
- aidd_cli-0.5.0/skill/templates/perf/plan-risk.test.sql +72 -0
- aidd_cli-0.5.0/skill/templates/perf/seed-generate-series.sql +52 -0
- aidd_cli-0.5.0/skill/templates/plan.md +62 -0
- aidd_cli-0.5.0/skill/templates/qa-audit.md +83 -0
- aidd_cli-0.5.0/skill/templates/research.md +8 -0
- aidd_cli-0.5.0/skill/templates/review-full.html +195 -0
- aidd_cli-0.5.0/skill/templates/review.html +1288 -0
- aidd_cli-0.5.0/skill/templates/spec.md +162 -0
- aidd_cli-0.5.0/skill/templates/tasks.md +107 -0
- aidd_cli-0.5.0/skill/templates/traceability.md +21 -0
- aidd_cli-0.5.0/skill/templates/visual-flow.toon +28 -0
- aidd_cli-0.1.0/aidd/cli.py +0 -359
- aidd_cli-0.1.0/aidd_cli.egg-info/SOURCES.txt +0 -25
- aidd_cli-0.1.0/tests/test_check_charter.py +0 -151
- aidd_cli-0.1.0/tests/test_check_spec.py +0 -129
- aidd_cli-0.1.0/tests/test_dot_aidd_mirror.py +0 -370
- aidd_cli-0.1.0/tests/test_extension_providers.py +0 -427
- aidd_cli-0.1.0/tests/test_find_spec.py +0 -239
- aidd_cli-0.1.0/tests/test_generate_adapters.py +0 -202
- aidd_cli-0.1.0/tests/test_hooks.py +0 -375
- aidd_cli-0.1.0/tests/test_install_ci.py +0 -108
- aidd_cli-0.1.0/tests/test_install_hooks.py +0 -81
- aidd_cli-0.1.0/tests/test_link_pr_to_task.py +0 -196
- aidd_cli-0.1.0/tests/test_marketplace.py +0 -280
- aidd_cli-0.1.0/tests/test_research_project.py +0 -110
- aidd_cli-0.1.0/tests/test_sync_issues.py +0 -167
- aidd_cli-0.1.0/tests/test_tasks_to_issues.py +0 -185
- aidd_cli-0.1.0/tests/test_validate_catalog_entry.py +0 -185
- {aidd_cli-0.1.0 → aidd_cli-0.5.0}/LICENSE +0 -0
- {aidd_cli-0.1.0 → aidd_cli-0.5.0}/aidd_cli.egg-info/dependency_links.txt +0 -0
- {aidd_cli-0.1.0 → aidd_cli-0.5.0}/aidd_cli.egg-info/entry_points.txt +0 -0
- {aidd_cli-0.1.0 → aidd_cli-0.5.0}/aidd_cli.egg-info/top_level.txt +0 -0
- {aidd_cli-0.1.0 → aidd_cli-0.5.0}/setup.cfg +0 -0
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
Metadata-Version: 2.4
|
|
2
2
|
Name: aidd-cli
|
|
3
|
-
Version: 0.
|
|
3
|
+
Version: 0.5.0
|
|
4
4
|
Summary: AIDD — a methodology for working with AI coding agents (CLI distribution)
|
|
5
5
|
License-Expression: MIT
|
|
6
6
|
Project-URL: Homepage, https://github.com/quimhen/aidd
|
|
@@ -23,12 +23,63 @@ Description-Content-Type: text/markdown
|
|
|
23
23
|
License-File: LICENSE
|
|
24
24
|
Dynamic: license-file
|
|
25
25
|
|
|
26
|
-
|
|
26
|
+
<div align="center">
|
|
27
27
|
|
|
28
|
-
|
|
29
|
-
|
|
30
|
-
|
|
31
|
-
|
|
28
|
+
# ⚡ AIDD — AI-Driven Development
|
|
29
|
+
### Stop AI coding agents from grading their own homework.
|
|
30
|
+
|
|
31
|
+
[](https://pypi.org/project/aidd-cli/)
|
|
32
|
+
[](LICENSE)
|
|
33
|
+
[](https://www.python.org/)
|
|
34
|
+
[](https://claude.ai)
|
|
35
|
+
[](https://github.com/sponsors/quimhen)
|
|
36
|
+
|
|
37
|
+
**A methodology with technical teeth for AI coding agents (Claude Code, OpenCode, Codex, Gemini CLI).**
|
|
38
|
+
It forces agents to search before creating, plan out loud before coding, and blocks self-approval
|
|
39
|
+
via automated hooks — not another AI coding agent itself, a set of habits your existing one follows.
|
|
40
|
+
|
|
41
|
+
[Explore Landing](https://aidd-landing.vercel.app/) · [Why AIDD](docs/WHY-AIDD.md) · [Pipeline Docs](docs/PIPELINE.md) · [Report Bug](https://github.com/quimhen/aidd/issues)
|
|
42
|
+
|
|
43
|
+
</div>
|
|
44
|
+
|
|
45
|
+
---
|
|
46
|
+
|
|
47
|
+
### 🚀 10-Second Quick Start
|
|
48
|
+
|
|
49
|
+
Get started in any existing repository without modifying your code:
|
|
50
|
+
|
|
51
|
+
```bash
|
|
52
|
+
# 1. Install the CLI
|
|
53
|
+
pip install aidd-cli
|
|
54
|
+
|
|
55
|
+
# 2. Initialize in your project
|
|
56
|
+
aidd init .
|
|
57
|
+
|
|
58
|
+
# 3. Check for architectural violations & spec gaps
|
|
59
|
+
aidd check-charter .
|
|
60
|
+
aidd search "auth flow"
|
|
61
|
+
```
|
|
62
|
+
|
|
63
|
+
> **Using Claude Code?** Install the technical enforcement gates with one command:
|
|
64
|
+
> ```bash
|
|
65
|
+
> cp -r skill ~/.claude/skills/aidd && python ~/.claude/skills/aidd/scripts/install_hooks.py
|
|
66
|
+
> ```
|
|
67
|
+
|
|
68
|
+
<div align="center">
|
|
69
|
+
<img src="docs/assets/demo-terminal.svg" alt="AIDD hook blocking a self-approved PR, then dispatching an independent QA subagent" width="760" />
|
|
70
|
+
</div>
|
|
71
|
+
|
|
72
|
+
---
|
|
73
|
+
|
|
74
|
+
### ⚖️ Why AIDD vs. traditional prompts / spec files
|
|
75
|
+
|
|
76
|
+
| Feature | Traditional rules (`CLAUDE.md`, `.cursorrules`) | AIDD methodology |
|
|
77
|
+
| :--- | :--- | :--- |
|
|
78
|
+
| **Compliance** | Optional (LLMs tend to skip long instructions) | **Strict / blocking** (real system hooks) |
|
|
79
|
+
| **Code review** | The same agent grades its own code | **Independent subagent audit, mandatory** |
|
|
80
|
+
| **UI mapping** | Duplicated, per-screen prose | **Unified codes** (`COMP-nnn`, `CTL-nnn`) |
|
|
81
|
+
| **Task tracking** | Disconnected markdown nobody re-checks | **Bidirectional sync** with GitHub / Azure DevOps / Bitbucket |
|
|
82
|
+
| **Context cost** | Dumps thousands of tokens into every prompt | **Indexed local lookup** (`find_spec.py`) |
|
|
32
83
|
|
|
33
84
|
## The problem
|
|
34
85
|
|
|
@@ -43,13 +94,46 @@ where the tool allows it, followed as a written discipline everywhere else.
|
|
|
43
94
|
- **Searches before creating.** Every request runs against a compact, auto-built graph of every
|
|
44
95
|
existing spec (`specs/index.toon`) before anything new gets created — a match means "amend this,"
|
|
45
96
|
not "start over."
|
|
97
|
+
- **Shows the plan before building it.** Step 1.5 turns each use case into an interactive
|
|
98
|
+
actors × processes flow with generated pseudocode (`aidd flow specs/001-login/visual-flow.toon
|
|
99
|
+
--open`) — the user clicks through it and corrects the flow by node, not in prose. Source is
|
|
100
|
+
a compact TOON file; layout, routing and pseudocode are derived, never hand-drawn.
|
|
46
101
|
- **Plans out loud.** Ambiguity gets resolved as an explicit question before code is written, not
|
|
47
102
|
discovered after a review.
|
|
48
103
|
- **Ships in small, reviewable pieces.** One code, one file, one PR — a wrong pass costs one edit,
|
|
49
104
|
not a rewrite.
|
|
50
105
|
- **Never lets the implementer grade its own work.** Step 6's review must come from a separate
|
|
51
|
-
agent dispatch. On Claude Code this is an actual technical
|
|
52
|
-
(`hooks/
|
|
106
|
+
agent dispatch, one per domain the change touched. On Claude Code this is an actual technical
|
|
107
|
+
gate (`hooks/rule_gate.py` + `stop_gate.py`), not just a rule stated in a doc.
|
|
108
|
+
- **Enforces the rules instead of asking nicely.** Hard rules R1–R12: estimates are agent
|
|
109
|
+
minutes per wave (human hours only as a reference), skipping a step needs the user's quoted
|
|
110
|
+
confirmation (verified against what the user typed or answered), the agent can't answer its own
|
|
111
|
+
Align questions (blank answers block planning, `repo —` sources must exist and cite a line),
|
|
112
|
+
plan/tasks can't be written out of order, `tasks.md` approval is the user's recorded "Approve"
|
|
113
|
+
answer plus a tamper-evident hash, a spec stays open until `aidd rules close`/`abandon` (touching
|
|
114
|
+
another spec can't switch the gates off), code edits are blocked for every open spec without a
|
|
115
|
+
recorded approval, every required domain needs its own distinct auditor, and a session can't
|
|
116
|
+
end on built code without its closing audits (R8 relaxes after 3 blocks); a ✅ screen/API/field
|
|
117
|
+
needs executed evidence, not a code reading (R10, freshness best-effort), a repeated bug needs a
|
|
118
|
+
root cause and pattern sweep (R11), and a task that reuses a screen/component must say whether it is a
|
|
119
|
+
new view or logic only (R12). Hooks write an
|
|
120
|
+
append-only evidence log (guarded against the agent by R9: the Write/Edit tools and a lexical shell guard, not a sandbox) (per-session in the temp dir, plus a project log only where
|
|
121
|
+
a `specs/` or `.aidd/` root exists) that the gates read; `aidd status` lists every open spec with the
|
|
122
|
+
ledger and **WHY blocked**, and `aidd rules check` prints `PASS|FAIL Rn … → fix`.
|
|
123
|
+
Honest limits: Claude Code only; **Bash and PowerShell can still edit code** and, through the CLI or
|
|
124
|
+
libraries, try to forge evidence (the guard is lexical, not a sandbox); the gates check that a
|
|
125
|
+
subagent ran and a question was answered, not that they were good; the user's click cannot be
|
|
126
|
+
proven; hooks fail open on a crash, launch failure or timeout (recorded as `hook_error` when
|
|
127
|
+
possible); session ids/resume behaviour are not verified. The rules stop accidental and
|
|
128
|
+
self-justified skipping, not a determined agent. `AIDD_RULES=off|0|false|no|warn` is the
|
|
129
|
+
owner's escape hatch.
|
|
130
|
+
- **Remembers the why.** Decisions, rejected options, bug root causes and constraints go into
|
|
131
|
+
`.aidd/memory/` as curated, code-anchored AIDD-TOON entries, committed with the code — no
|
|
132
|
+
daemon, no LLM observer, no vectors, stdlib only. Search is progressive: `aidd mem search`
|
|
133
|
+
returns one line per hit, `aidd mem show <id>` the full row. On Claude Code `memory_context.py` surfaces
|
|
134
|
+
the relevant entries at session start (installed by default; the per-file `memory_file_context.py` is opt-in). `STATE.md` stays
|
|
135
|
+
"where am I", `specs/index.toon` stays "what exists"; memory is "why". A one-off, read-only
|
|
136
|
+
importer can seed it from a claude-mem database (secrets and IPs scrubbed). Memory is committed content that gets injected into sessions: review it like code.
|
|
53
137
|
- **Works the same across tools.** One methodology (`skill/AIDD.md`), thin adapters per tool — no
|
|
54
138
|
relearning the process when the assistant changes. For Claude Code, OpenCode, and Codex, that's
|
|
55
139
|
a native skill/AGENTS.md pointer; for Gemini CLI, Cursor, Windsurf, Cline, and GitHub Copilot,
|
|
@@ -89,9 +173,11 @@ where the tool allows it, followed as a written discipline everywhere else.
|
|
|
89
173
|
|
|
90
174
|
## What's in this repo
|
|
91
175
|
|
|
92
|
-
- **`skill/`** — the Claude Code skill: `SKILL.md`, enforcement hooks (`hooks
|
|
176
|
+
- **`skill/`** — the Claude Code skill: `SKILL.md`, enforcement hooks (`hooks/`: ten by default, centred on the single `rule_gate.py`), and every
|
|
93
177
|
script/template (`scripts/`, `templates/`). `skill/AIDD.md` is the tool-agnostic methodology
|
|
94
|
-
core every other adapter — including this repo's own CLI — points back to.
|
|
178
|
+
core every other adapter — including this repo's own CLI — points back to. Includes AIDD Memory
|
|
179
|
+
(`scripts/aidd_memory.py`, `aidd_memory_import.py`; hooks `memory_context.py`,
|
|
180
|
+
`memory_file_context.py`, opt-in via `install_hooks.py --with-memory-file-hook`).
|
|
95
181
|
- **`aidd/`** — the CLI (`pip install -e .` → the `aidd` command): a thin dispatcher over
|
|
96
182
|
`skill/scripts/`, for a human at a terminal or a CI job, no AI agent required.
|
|
97
183
|
- **`commands/`** — the eight `/aidd-*` pipeline-stage commands for Claude Code.
|
|
@@ -118,12 +204,27 @@ python ~/.claude/skills/aidd/scripts/install_hooks.py
|
|
|
118
204
|
## Install (the CLI — for a terminal or CI, no AI agent needed)
|
|
119
205
|
|
|
120
206
|
```bash
|
|
121
|
-
|
|
122
|
-
pip install -e .
|
|
207
|
+
pip install aidd-cli
|
|
123
208
|
aidd init /path/to/your/project # installs the .aidd/ bundle there
|
|
124
209
|
aidd search "login" # search the spec graph
|
|
125
210
|
aidd check specs/001-login/ # mechanical gap-check
|
|
126
211
|
aidd check-charter . # run charter.md's checkable rules
|
|
212
|
+
aidd status # hard-rules ledger of all open specs + WHY blocked (--json; exit 0 always)
|
|
213
|
+
aidd rules check specs/001-login/ # PASS|FAIL Rn message -> fix (exit 1 on a violation)
|
|
214
|
+
aidd rules approve specs/001-login/ # write the tasks.md Approved: line (needs the user's recorded "Approve" answer)
|
|
215
|
+
aidd rules close 001-login # complete the spec (needs approval + qa-audit + auditors + the user's "Yes, close" answer)
|
|
216
|
+
aidd rules abandon 001-login # drop it (needs the user's recorded "Abandon" answer)
|
|
217
|
+
aidd flow specs/001-login/visual-flow.toon --open # Flowmap: interactive flow + pseudocode
|
|
218
|
+
aidd mem add --type decision --title "Login uses magic links" --why "No password storage" --codes US-001,API-003 --scope 001-login
|
|
219
|
+
aidd mem search login # one line per hit: id, date, type, codes, title
|
|
220
|
+
aidd mem show m-3fa91c02 # full entry (why, files, supersedes)
|
|
221
|
+
aidd mem timeline m-3fa91c02 # neighbouring entries of the same scope, by date
|
|
222
|
+
aidd mem file skill/scripts/flowmap.py # entries that mention a file
|
|
223
|
+
aidd mem inject # the digest the SessionStart hook injects (also: inject --file PATH)
|
|
224
|
+
aidd mem stats # counts by type/scope, bytes
|
|
225
|
+
aidd mem compact --before 2026-07-01 --apply # archive superseded/old entries (dry run without --apply)
|
|
226
|
+
aidd mem import-claude-mem <claude-mem.db> --project myproj --since 2026-01-01 --dry-run # one-off, read-only import (also --scope, --include, --no-summaries)
|
|
227
|
+
# AIDD_MEMORY_DIR overrides .aidd/memory/; `aidd mem --root X <cmd>` works before or after the command; `add --date` backdates
|
|
127
228
|
aidd tasks-to-issues specs/001-login/tasks.md --apply # sync tasks (github by default)
|
|
128
229
|
aidd tasks-to-issues specs/001-login/tasks.md --provider azure_devops --org ... --project ... --apply
|
|
129
230
|
aidd adapters generate all /path/to/your/project # native commands for Gemini CLI, Cursor, Windsurf, Cline, Copilot
|
|
@@ -135,7 +236,13 @@ aidd tracker link-pr specs/001-login/tasks.md --pr-url ... --branch ... --apply
|
|
|
135
236
|
aidd ci install github /path/to/your/project # install the GitHub Actions CI template
|
|
136
237
|
```
|
|
137
238
|
|
|
138
|
-
|
|
239
|
+
Published on PyPI as [`aidd-cli`](https://pypi.org/project/aidd-cli/). To work on AIDD itself
|
|
240
|
+
instead of just using it, clone and install in editable mode:
|
|
241
|
+
|
|
242
|
+
```bash
|
|
243
|
+
git clone https://github.com/quimhen/aidd.git && cd aidd
|
|
244
|
+
pip install -e .
|
|
245
|
+
```
|
|
139
246
|
|
|
140
247
|
## Install (other tools)
|
|
141
248
|
|
|
@@ -160,8 +267,8 @@ See `adapters/README.md`.
|
|
|
160
267
|
Early — the core pipeline, the search/graph tools, the charter checker, the graph-coherence
|
|
161
268
|
gate, the issue-tracker sync (GitHub/Azure DevOps/Bitbucket, plus bidirectional status sync
|
|
162
269
|
via `aidd tracker sync` and PR<->task linking via `aidd tracker link-pr`), CI templates and
|
|
163
|
-
`aidd ci install`, the Claude Code hooks (
|
|
164
|
-
graph-coherence gates), the multi-agent adapter generator (`generate_adapters.py`, rendering
|
|
270
|
+
`aidd ci install`, the Claude Code hooks (ten by default, plus an opt-in per-file memory hook; one `rule_gate.py` enforcing hard rules R1–R12
|
|
271
|
+
including the independent-audit and graph-coherence gates, and a `stop_gate.py`), the evidence log, `aidd status`/`aidd rules`, the multi-agent adapter generator (`generate_adapters.py`, rendering
|
|
165
272
|
Gemini CLI/Cursor/Windsurf/Cline/Copilot commands from one source), the real auto-discovery
|
|
166
273
|
extension registry (`extension_registry.py`, loading provider/adapter/hook extensions from
|
|
167
274
|
`skill/extensions/**/manifest.json` and `.aidd/extensions/**/manifest.json` — see
|
|
@@ -169,11 +276,16 @@ extension registry (`extension_registry.py`, loading provider/adapter/hook exten
|
|
|
169
276
|
`validate_catalog_entry.py`, one first-party package shipped so far: `hotfix-report`) are built,
|
|
170
277
|
tested (stdlib `unittest`, run in CI), and have been run against a real production
|
|
171
278
|
codebase. OpenCode, Codex, and the five generated adapter targets follow the same methodology;
|
|
172
|
-
only the Claude Code hooks currently give it technical teeth.
|
|
173
|
-
marketplace is intentionally not a fully automated accept path: CI validates structure and
|
|
279
|
+
only the Claude Code hooks currently give it technical teeth. Published on PyPI as `aidd-cli`.
|
|
280
|
+
The marketplace is intentionally not a fully automated accept path: CI validates structure and
|
|
174
281
|
flags anything worth a human's attention, but a maintainer reviews and merges every catalog
|
|
175
282
|
submission by hand — see `catalog/README.md`.
|
|
176
283
|
|
|
284
|
+
## Support this project
|
|
285
|
+
|
|
286
|
+
If AIDD saves your agent sessions real time, [sponsor it on GitHub](https://github.com/sponsors/quimhen)
|
|
287
|
+
— it funds the time spent maintaining the pipeline, the hooks, and the multi-agent adapters.
|
|
288
|
+
|
|
177
289
|
## License
|
|
178
290
|
|
|
179
291
|
MIT — see [LICENSE](LICENSE).
|
|
@@ -1,9 +1,60 @@
|
|
|
1
|
-
|
|
1
|
+
<div align="center">
|
|
2
2
|
|
|
3
|
-
|
|
4
|
-
|
|
5
|
-
|
|
6
|
-
|
|
3
|
+
# ⚡ AIDD — AI-Driven Development
|
|
4
|
+
### Stop AI coding agents from grading their own homework.
|
|
5
|
+
|
|
6
|
+
[](https://pypi.org/project/aidd-cli/)
|
|
7
|
+
[](LICENSE)
|
|
8
|
+
[](https://www.python.org/)
|
|
9
|
+
[](https://claude.ai)
|
|
10
|
+
[](https://github.com/sponsors/quimhen)
|
|
11
|
+
|
|
12
|
+
**A methodology with technical teeth for AI coding agents (Claude Code, OpenCode, Codex, Gemini CLI).**
|
|
13
|
+
It forces agents to search before creating, plan out loud before coding, and blocks self-approval
|
|
14
|
+
via automated hooks — not another AI coding agent itself, a set of habits your existing one follows.
|
|
15
|
+
|
|
16
|
+
[Explore Landing](https://aidd-landing.vercel.app/) · [Why AIDD](docs/WHY-AIDD.md) · [Pipeline Docs](docs/PIPELINE.md) · [Report Bug](https://github.com/quimhen/aidd/issues)
|
|
17
|
+
|
|
18
|
+
</div>
|
|
19
|
+
|
|
20
|
+
---
|
|
21
|
+
|
|
22
|
+
### 🚀 10-Second Quick Start
|
|
23
|
+
|
|
24
|
+
Get started in any existing repository without modifying your code:
|
|
25
|
+
|
|
26
|
+
```bash
|
|
27
|
+
# 1. Install the CLI
|
|
28
|
+
pip install aidd-cli
|
|
29
|
+
|
|
30
|
+
# 2. Initialize in your project
|
|
31
|
+
aidd init .
|
|
32
|
+
|
|
33
|
+
# 3. Check for architectural violations & spec gaps
|
|
34
|
+
aidd check-charter .
|
|
35
|
+
aidd search "auth flow"
|
|
36
|
+
```
|
|
37
|
+
|
|
38
|
+
> **Using Claude Code?** Install the technical enforcement gates with one command:
|
|
39
|
+
> ```bash
|
|
40
|
+
> cp -r skill ~/.claude/skills/aidd && python ~/.claude/skills/aidd/scripts/install_hooks.py
|
|
41
|
+
> ```
|
|
42
|
+
|
|
43
|
+
<div align="center">
|
|
44
|
+
<img src="docs/assets/demo-terminal.svg" alt="AIDD hook blocking a self-approved PR, then dispatching an independent QA subagent" width="760" />
|
|
45
|
+
</div>
|
|
46
|
+
|
|
47
|
+
---
|
|
48
|
+
|
|
49
|
+
### ⚖️ Why AIDD vs. traditional prompts / spec files
|
|
50
|
+
|
|
51
|
+
| Feature | Traditional rules (`CLAUDE.md`, `.cursorrules`) | AIDD methodology |
|
|
52
|
+
| :--- | :--- | :--- |
|
|
53
|
+
| **Compliance** | Optional (LLMs tend to skip long instructions) | **Strict / blocking** (real system hooks) |
|
|
54
|
+
| **Code review** | The same agent grades its own code | **Independent subagent audit, mandatory** |
|
|
55
|
+
| **UI mapping** | Duplicated, per-screen prose | **Unified codes** (`COMP-nnn`, `CTL-nnn`) |
|
|
56
|
+
| **Task tracking** | Disconnected markdown nobody re-checks | **Bidirectional sync** with GitHub / Azure DevOps / Bitbucket |
|
|
57
|
+
| **Context cost** | Dumps thousands of tokens into every prompt | **Indexed local lookup** (`find_spec.py`) |
|
|
7
58
|
|
|
8
59
|
## The problem
|
|
9
60
|
|
|
@@ -18,13 +69,46 @@ where the tool allows it, followed as a written discipline everywhere else.
|
|
|
18
69
|
- **Searches before creating.** Every request runs against a compact, auto-built graph of every
|
|
19
70
|
existing spec (`specs/index.toon`) before anything new gets created — a match means "amend this,"
|
|
20
71
|
not "start over."
|
|
72
|
+
- **Shows the plan before building it.** Step 1.5 turns each use case into an interactive
|
|
73
|
+
actors × processes flow with generated pseudocode (`aidd flow specs/001-login/visual-flow.toon
|
|
74
|
+
--open`) — the user clicks through it and corrects the flow by node, not in prose. Source is
|
|
75
|
+
a compact TOON file; layout, routing and pseudocode are derived, never hand-drawn.
|
|
21
76
|
- **Plans out loud.** Ambiguity gets resolved as an explicit question before code is written, not
|
|
22
77
|
discovered after a review.
|
|
23
78
|
- **Ships in small, reviewable pieces.** One code, one file, one PR — a wrong pass costs one edit,
|
|
24
79
|
not a rewrite.
|
|
25
80
|
- **Never lets the implementer grade its own work.** Step 6's review must come from a separate
|
|
26
|
-
agent dispatch. On Claude Code this is an actual technical
|
|
27
|
-
(`hooks/
|
|
81
|
+
agent dispatch, one per domain the change touched. On Claude Code this is an actual technical
|
|
82
|
+
gate (`hooks/rule_gate.py` + `stop_gate.py`), not just a rule stated in a doc.
|
|
83
|
+
- **Enforces the rules instead of asking nicely.** Hard rules R1–R12: estimates are agent
|
|
84
|
+
minutes per wave (human hours only as a reference), skipping a step needs the user's quoted
|
|
85
|
+
confirmation (verified against what the user typed or answered), the agent can't answer its own
|
|
86
|
+
Align questions (blank answers block planning, `repo —` sources must exist and cite a line),
|
|
87
|
+
plan/tasks can't be written out of order, `tasks.md` approval is the user's recorded "Approve"
|
|
88
|
+
answer plus a tamper-evident hash, a spec stays open until `aidd rules close`/`abandon` (touching
|
|
89
|
+
another spec can't switch the gates off), code edits are blocked for every open spec without a
|
|
90
|
+
recorded approval, every required domain needs its own distinct auditor, and a session can't
|
|
91
|
+
end on built code without its closing audits (R8 relaxes after 3 blocks); a ✅ screen/API/field
|
|
92
|
+
needs executed evidence, not a code reading (R10, freshness best-effort), a repeated bug needs a
|
|
93
|
+
root cause and pattern sweep (R11), and a task that reuses a screen/component must say whether it is a
|
|
94
|
+
new view or logic only (R12). Hooks write an
|
|
95
|
+
append-only evidence log (guarded against the agent by R9: the Write/Edit tools and a lexical shell guard, not a sandbox) (per-session in the temp dir, plus a project log only where
|
|
96
|
+
a `specs/` or `.aidd/` root exists) that the gates read; `aidd status` lists every open spec with the
|
|
97
|
+
ledger and **WHY blocked**, and `aidd rules check` prints `PASS|FAIL Rn … → fix`.
|
|
98
|
+
Honest limits: Claude Code only; **Bash and PowerShell can still edit code** and, through the CLI or
|
|
99
|
+
libraries, try to forge evidence (the guard is lexical, not a sandbox); the gates check that a
|
|
100
|
+
subagent ran and a question was answered, not that they were good; the user's click cannot be
|
|
101
|
+
proven; hooks fail open on a crash, launch failure or timeout (recorded as `hook_error` when
|
|
102
|
+
possible); session ids/resume behaviour are not verified. The rules stop accidental and
|
|
103
|
+
self-justified skipping, not a determined agent. `AIDD_RULES=off|0|false|no|warn` is the
|
|
104
|
+
owner's escape hatch.
|
|
105
|
+
- **Remembers the why.** Decisions, rejected options, bug root causes and constraints go into
|
|
106
|
+
`.aidd/memory/` as curated, code-anchored AIDD-TOON entries, committed with the code — no
|
|
107
|
+
daemon, no LLM observer, no vectors, stdlib only. Search is progressive: `aidd mem search`
|
|
108
|
+
returns one line per hit, `aidd mem show <id>` the full row. On Claude Code `memory_context.py` surfaces
|
|
109
|
+
the relevant entries at session start (installed by default; the per-file `memory_file_context.py` is opt-in). `STATE.md` stays
|
|
110
|
+
"where am I", `specs/index.toon` stays "what exists"; memory is "why". A one-off, read-only
|
|
111
|
+
importer can seed it from a claude-mem database (secrets and IPs scrubbed). Memory is committed content that gets injected into sessions: review it like code.
|
|
28
112
|
- **Works the same across tools.** One methodology (`skill/AIDD.md`), thin adapters per tool — no
|
|
29
113
|
relearning the process when the assistant changes. For Claude Code, OpenCode, and Codex, that's
|
|
30
114
|
a native skill/AGENTS.md pointer; for Gemini CLI, Cursor, Windsurf, Cline, and GitHub Copilot,
|
|
@@ -64,9 +148,11 @@ where the tool allows it, followed as a written discipline everywhere else.
|
|
|
64
148
|
|
|
65
149
|
## What's in this repo
|
|
66
150
|
|
|
67
|
-
- **`skill/`** — the Claude Code skill: `SKILL.md`, enforcement hooks (`hooks
|
|
151
|
+
- **`skill/`** — the Claude Code skill: `SKILL.md`, enforcement hooks (`hooks/`: ten by default, centred on the single `rule_gate.py`), and every
|
|
68
152
|
script/template (`scripts/`, `templates/`). `skill/AIDD.md` is the tool-agnostic methodology
|
|
69
|
-
core every other adapter — including this repo's own CLI — points back to.
|
|
153
|
+
core every other adapter — including this repo's own CLI — points back to. Includes AIDD Memory
|
|
154
|
+
(`scripts/aidd_memory.py`, `aidd_memory_import.py`; hooks `memory_context.py`,
|
|
155
|
+
`memory_file_context.py`, opt-in via `install_hooks.py --with-memory-file-hook`).
|
|
70
156
|
- **`aidd/`** — the CLI (`pip install -e .` → the `aidd` command): a thin dispatcher over
|
|
71
157
|
`skill/scripts/`, for a human at a terminal or a CI job, no AI agent required.
|
|
72
158
|
- **`commands/`** — the eight `/aidd-*` pipeline-stage commands for Claude Code.
|
|
@@ -93,12 +179,27 @@ python ~/.claude/skills/aidd/scripts/install_hooks.py
|
|
|
93
179
|
## Install (the CLI — for a terminal or CI, no AI agent needed)
|
|
94
180
|
|
|
95
181
|
```bash
|
|
96
|
-
|
|
97
|
-
pip install -e .
|
|
182
|
+
pip install aidd-cli
|
|
98
183
|
aidd init /path/to/your/project # installs the .aidd/ bundle there
|
|
99
184
|
aidd search "login" # search the spec graph
|
|
100
185
|
aidd check specs/001-login/ # mechanical gap-check
|
|
101
186
|
aidd check-charter . # run charter.md's checkable rules
|
|
187
|
+
aidd status # hard-rules ledger of all open specs + WHY blocked (--json; exit 0 always)
|
|
188
|
+
aidd rules check specs/001-login/ # PASS|FAIL Rn message -> fix (exit 1 on a violation)
|
|
189
|
+
aidd rules approve specs/001-login/ # write the tasks.md Approved: line (needs the user's recorded "Approve" answer)
|
|
190
|
+
aidd rules close 001-login # complete the spec (needs approval + qa-audit + auditors + the user's "Yes, close" answer)
|
|
191
|
+
aidd rules abandon 001-login # drop it (needs the user's recorded "Abandon" answer)
|
|
192
|
+
aidd flow specs/001-login/visual-flow.toon --open # Flowmap: interactive flow + pseudocode
|
|
193
|
+
aidd mem add --type decision --title "Login uses magic links" --why "No password storage" --codes US-001,API-003 --scope 001-login
|
|
194
|
+
aidd mem search login # one line per hit: id, date, type, codes, title
|
|
195
|
+
aidd mem show m-3fa91c02 # full entry (why, files, supersedes)
|
|
196
|
+
aidd mem timeline m-3fa91c02 # neighbouring entries of the same scope, by date
|
|
197
|
+
aidd mem file skill/scripts/flowmap.py # entries that mention a file
|
|
198
|
+
aidd mem inject # the digest the SessionStart hook injects (also: inject --file PATH)
|
|
199
|
+
aidd mem stats # counts by type/scope, bytes
|
|
200
|
+
aidd mem compact --before 2026-07-01 --apply # archive superseded/old entries (dry run without --apply)
|
|
201
|
+
aidd mem import-claude-mem <claude-mem.db> --project myproj --since 2026-01-01 --dry-run # one-off, read-only import (also --scope, --include, --no-summaries)
|
|
202
|
+
# AIDD_MEMORY_DIR overrides .aidd/memory/; `aidd mem --root X <cmd>` works before or after the command; `add --date` backdates
|
|
102
203
|
aidd tasks-to-issues specs/001-login/tasks.md --apply # sync tasks (github by default)
|
|
103
204
|
aidd tasks-to-issues specs/001-login/tasks.md --provider azure_devops --org ... --project ... --apply
|
|
104
205
|
aidd adapters generate all /path/to/your/project # native commands for Gemini CLI, Cursor, Windsurf, Cline, Copilot
|
|
@@ -110,7 +211,13 @@ aidd tracker link-pr specs/001-login/tasks.md --pr-url ... --branch ... --apply
|
|
|
110
211
|
aidd ci install github /path/to/your/project # install the GitHub Actions CI template
|
|
111
212
|
```
|
|
112
213
|
|
|
113
|
-
|
|
214
|
+
Published on PyPI as [`aidd-cli`](https://pypi.org/project/aidd-cli/). To work on AIDD itself
|
|
215
|
+
instead of just using it, clone and install in editable mode:
|
|
216
|
+
|
|
217
|
+
```bash
|
|
218
|
+
git clone https://github.com/quimhen/aidd.git && cd aidd
|
|
219
|
+
pip install -e .
|
|
220
|
+
```
|
|
114
221
|
|
|
115
222
|
## Install (other tools)
|
|
116
223
|
|
|
@@ -135,8 +242,8 @@ See `adapters/README.md`.
|
|
|
135
242
|
Early — the core pipeline, the search/graph tools, the charter checker, the graph-coherence
|
|
136
243
|
gate, the issue-tracker sync (GitHub/Azure DevOps/Bitbucket, plus bidirectional status sync
|
|
137
244
|
via `aidd tracker sync` and PR<->task linking via `aidd tracker link-pr`), CI templates and
|
|
138
|
-
`aidd ci install`, the Claude Code hooks (
|
|
139
|
-
graph-coherence gates), the multi-agent adapter generator (`generate_adapters.py`, rendering
|
|
245
|
+
`aidd ci install`, the Claude Code hooks (ten by default, plus an opt-in per-file memory hook; one `rule_gate.py` enforcing hard rules R1–R12
|
|
246
|
+
including the independent-audit and graph-coherence gates, and a `stop_gate.py`), the evidence log, `aidd status`/`aidd rules`, the multi-agent adapter generator (`generate_adapters.py`, rendering
|
|
140
247
|
Gemini CLI/Cursor/Windsurf/Cline/Copilot commands from one source), the real auto-discovery
|
|
141
248
|
extension registry (`extension_registry.py`, loading provider/adapter/hook extensions from
|
|
142
249
|
`skill/extensions/**/manifest.json` and `.aidd/extensions/**/manifest.json` — see
|
|
@@ -144,11 +251,16 @@ extension registry (`extension_registry.py`, loading provider/adapter/hook exten
|
|
|
144
251
|
`validate_catalog_entry.py`, one first-party package shipped so far: `hotfix-report`) are built,
|
|
145
252
|
tested (stdlib `unittest`, run in CI), and have been run against a real production
|
|
146
253
|
codebase. OpenCode, Codex, and the five generated adapter targets follow the same methodology;
|
|
147
|
-
only the Claude Code hooks currently give it technical teeth.
|
|
148
|
-
marketplace is intentionally not a fully automated accept path: CI validates structure and
|
|
254
|
+
only the Claude Code hooks currently give it technical teeth. Published on PyPI as `aidd-cli`.
|
|
255
|
+
The marketplace is intentionally not a fully automated accept path: CI validates structure and
|
|
149
256
|
flags anything worth a human's attention, but a maintainer reviews and merges every catalog
|
|
150
257
|
submission by hand — see `catalog/README.md`.
|
|
151
258
|
|
|
259
|
+
## Support this project
|
|
260
|
+
|
|
261
|
+
If AIDD saves your agent sessions real time, [sponsor it on GitHub](https://github.com/sponsors/quimhen)
|
|
262
|
+
— it funds the time spent maintaining the pipeline, the hooks, and the multi-agent adapters.
|
|
263
|
+
|
|
152
264
|
## License
|
|
153
265
|
|
|
154
266
|
MIT — see [LICENSE](LICENSE).
|
|
@@ -0,0 +1,29 @@
|
|
|
1
|
+
## AIDD
|
|
2
|
+
|
|
3
|
+
This project uses AIDD (AI-Driven Development) as its default pipeline for **any** change —
|
|
4
|
+
UI or backend-only, a new feature or a one-line fix, regardless of size. Full methodology:
|
|
5
|
+
`.aidd/AIDD.md` (~120 KB: hosts truncate it, so do NOT read it whole). Read `.aidd/CORE.md`
|
|
6
|
+
(~6 KB) before planning or implementing a requirement, change, modification, update, fix, or
|
|
7
|
+
improvement — not only when a mockup is mentioned. It says how to check the hard rules when no
|
|
8
|
+
hook does it for you, what "done" requires (executed, and compared with the mockup for any
|
|
9
|
+
screen), and how to read one section of `AIDD.md` on demand. Check the install with `aidd doctor`
|
|
10
|
+
(fix with `aidd install --host gemini|opencode`).
|
|
11
|
+
|
|
12
|
+
**Before creating anything new, search first — this is not optional:**
|
|
13
|
+
```
|
|
14
|
+
python .aidd/scripts/find_spec.py <a few keywords describing the request, or a SCREEN-XX/CTL-nnn/COMP-nnn/API-nnn code>
|
|
15
|
+
```
|
|
16
|
+
- Exit code 0 → inspect the top candidate and relevant sources before classifying it as an amendment; ranking is not proof of a match.
|
|
17
|
+
- Exit code 1 → no candidate was found; this does not prove absence. Check relevant specs and naming variants before creating a new spec.
|
|
18
|
+
- `python .aidd/scripts/find_spec.py --tree <spec-id>` prints that spec's real use-case → screen → component → control → API graph.
|
|
19
|
+
- `python .aidd/scripts/check_spec.py specs/[###-feature]/` is a mechanical gap-checker — run it before a manual audit pass (Step 6).
|
|
20
|
+
|
|
21
|
+
State classification out loud before proceeding (`AIDD intake: NEW SPEC | AMENDMENT | AMENDMENT — Fast Lane — <reason>`), per `.aidd/AIDD.md`'s Step -1.
|
|
22
|
+
|
|
23
|
+
Rules:
|
|
24
|
+
- Never skip straight to writing code for a requirement/change — go through `.aidd/AIDD.md`'s pipeline (search → align → plan → tasks → build → converge), using its Fast Lane for a genuinely small, contained change.
|
|
25
|
+
- A change with no visual surface (backend, API, DB, infra) skips the visual-only steps (0/1/1.5) but still goes through the rest — AIDD is not UI-only.
|
|
26
|
+
- Shared-First & Component-Driven Development (CDD): eradicate code duplication (DRY), assume a shared component library/monorepo, favor Inversion of Control (Compound Components, slots), and structure UI requests in 3 sections: [AUDITORÍA DE REUTILIZACIÓN], [PROPUESTA DE COMPONENTE COMPARTIDO], [IMPLEMENTACIÓN LOCAL / EXTENSIÓN].
|
|
27
|
+
- Templates for every artifact live in `.aidd/templates/`; copy the matching one for whatever step you're on instead of hand-writing table headers from scratch.
|
|
28
|
+
- `specs/index.toon` is an auto-built, auto-invalidated search index — never hand-edit it; `find_spec.py` rebuilds it itself when a spec's files change.
|
|
29
|
+
- Verify several open specs in ONE pass (E12): `aidd verify --round <spec>... --dry-run` (plan and savings), freeze the code, `aidd verify --round <spec>...` (rows their own spec already passed on the same Scope, code, Inputs and environment identity are `REUSED`, not run — never a FAIL, never on an uncomputable fingerprint, never with `--full` or `Reuse: never`; every other identical row runs once, shared only on an unchanged environment fingerprint), then one round of auditors per spec. Spec by spec only when a single spec is ready.
|
|
@@ -0,0 +1,16 @@
|
|
|
1
|
+
# AIDD
|
|
2
|
+
|
|
3
|
+
This project uses AIDD (AI-Driven Development) as its default pipeline for any change: UI or
|
|
4
|
+
backend-only, a new feature or a one-line fix.
|
|
5
|
+
|
|
6
|
+
@.aidd/CORE.md
|
|
7
|
+
|
|
8
|
+
The line above imports the short core (host detection, how to check the hard rules without hooks,
|
|
9
|
+
the pipeline, what "done" means). Read it before planning or editing anything. Do NOT read
|
|
10
|
+
`.aidd/AIDD.md` whole: it is ~120 KB and gets truncated; `CORE.md` section 6 shows how to read
|
|
11
|
+
only the section you need.
|
|
12
|
+
|
|
13
|
+
Gate missing or unsure? `aidd doctor --host gemini`; to install it: `aidd install --host gemini`.
|
|
14
|
+
|
|
15
|
+
Before creating anything new: `python .aidd/scripts/find_spec.py <keywords>` and state
|
|
16
|
+
`AIDD intake: NEW SPEC | AMENDMENT | AMENDMENT — Fast Lane — <reason>`.
|