pan-wizard 3.35.0 → 3.37.0
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.
- package/CHANGELOG.md +3645 -0
- package/README.md +46 -52
- package/agents/pan-conductor.md +2 -3
- package/agents/pan-counterfactual.md +1 -2
- package/agents/pan-debugger.md +14 -3
- package/agents/pan-design-checker.md +1 -2
- package/agents/pan-designer.md +2 -3
- package/agents/pan-document_code.md +1 -2
- package/agents/pan-executor.md +25 -0
- package/agents/pan-hardener.md +4 -3
- package/agents/pan-integration-checker.md +1 -2
- package/agents/pan-knowledge.md +1 -2
- package/agents/pan-meta-reviewer.md +3 -3
- package/agents/pan-optimizer.md +4 -4
- package/agents/pan-phase-researcher.md +80 -475
- package/agents/pan-plan-checker.md +40 -3
- package/agents/pan-planner.md +21 -7
- package/agents/pan-previewer.md +3 -2
- package/agents/pan-project-researcher.md +72 -493
- package/agents/pan-release.md +1 -1
- package/agents/pan-reviewer.md +14 -3
- package/agents/pan-roadmapper.md +12 -7
- package/agents/pan-verifier.md +9 -4
- package/bin/install-lib.cjs +134 -14
- package/bin/install.js +59 -12
- package/commands/pan/army.md +1 -1
- package/commands/pan/autodev.md +76 -0
- package/commands/pan/cost.md +21 -1
- package/commands/pan/exec-phase.md +3 -4
- package/commands/pan/focus-auto.md +14 -7
- package/commands/pan/focus-design.md +4 -5
- package/commands/pan/focus-docs.md +24 -0
- package/commands/pan/learn.md +13 -6
- package/commands/pan/milestone-audit.md +8 -1
- package/commands/pan/milestone-done.md +10 -5
- package/commands/pan/milestone-new.md +2 -2
- package/commands/pan/new-project.md +15 -12
- package/commands/pan/plan-phase.md +15 -9
- package/commands/pan/preview.md +1 -1
- package/commands/pan/quick.md +1 -1
- package/commands/pan/review-deep.md +11 -5
- package/commands/pan/settings.md +6 -0
- package/commands/pan/verify-phase.md +6 -2
- package/docs/AGENTS.md +917 -0
- package/docs/ARCHITECTURE.md +1054 -0
- package/docs/CLI-REFERENCE.md +4582 -0
- package/docs/COMPARISON.md +116 -0
- package/docs/DEVELOPMENT.md +490 -0
- package/docs/ECOSYSTEM-REVIEW-2026-06.md +107 -0
- package/docs/ECOSYSTEM-REVIEW-2026-08.md +349 -0
- package/docs/EXAMPLES.md +584 -0
- package/docs/FAQ.md +254 -0
- package/docs/FIELD-HARVEST-2026-07.md +88 -0
- package/docs/FIELD-REPORT-army-2026-06.md +121 -0
- package/docs/HOOKS.md +638 -0
- package/docs/IMPROVEMENT-TODO.md +246 -0
- package/docs/INTERNALS.md +485 -0
- package/docs/MIGRATION-v2-to-v3.md +258 -0
- package/docs/SECURITY-HARDENING.md +230 -0
- package/docs/SKILLS-ENHANCEMENT-PLAN-V2.md +122 -0
- package/docs/SKILLS-FULL-TEXT.md +9685 -0
- package/docs/SKILLS-REFERENCE.md +1037 -0
- package/docs/TROUBLESHOOTING.md +1084 -0
- package/docs/USER-GUIDE.md +1437 -0
- package/docs/archive/SKILLS-ENHANCEMENT-PLAN.md +230 -0
- package/docs/audits/AUDIT-2026-08-e2e-ROADMAP.md +117 -0
- package/docs/audits/AUDIT-2026-08-e2e-VERIFICATION.md +693 -0
- package/docs/audits/AUDIT-2026-08-e2e.md +1546 -0
- package/docs/audits/OUTSTANDING-2026-08.md +275 -0
- package/docs/audits/STABILITY-TEST-2026-08.md +640 -0
- package/docs/branding/README-brand-draft.md +119 -0
- package/docs/branding/handoff/IMPLEMENTATION.md +205 -0
- package/docs/branding/handoff/README-brand-draft.md +119 -0
- package/docs/branding/handoff/image-prompts.md +46 -0
- package/docs/branding/image-prompts.md +46 -0
- package/docs/branding/nano-banana-prompts.txt +60 -0
- package/docs/context-monitor.md +13 -0
- package/docs/decisions/ADR-0001-strategic-feature-prioritization.md +83 -0
- package/docs/decisions/ADR-0002-copilot-cli-runtime.md +71 -0
- package/docs/decisions/ADR-0003-smart-execution.md +84 -0
- package/docs/decisions/ADR-0004-copilot-cli-interaction.md +89 -0
- package/docs/decisions/ADR-0005-command-naming.md +80 -0
- package/docs/decisions/ADR-0006-focus-commands.md +101 -0
- package/docs/decisions/ADR-0007-ship-v1.md +73 -0
- package/docs/decisions/ADR-0008-map-codebase-mermaid-togaf.md +80 -0
- package/docs/decisions/ADR-0009-production-deployment-checklist.md +68 -0
- package/docs/decisions/ADR-0010-standards-integration.md +79 -0
- package/docs/decisions/ADR-0011-help-commands-rewrite.md +64 -0
- package/docs/decisions/ADR-0012-contributors-attribution-index.md +42 -0
- package/docs/decisions/ADR-0013-strategic-feature-additions.md +53 -0
- package/docs/decisions/ADR-0014-internal-cleanup-code-quality.md +53 -0
- package/docs/decisions/ADR-0015-focus-auto-runner.md +296 -0
- package/docs/decisions/ADR-0016-enhanced-doc-sync.md +54 -0
- package/docs/decisions/ADR-0017-enhanced-focus-design.md +69 -0
- package/docs/decisions/ADR-0018-user-reality-testing.md +84 -0
- package/docs/decisions/ADR-0019-end-to-end-user-acceptance-testing.md +113 -0
- package/docs/decisions/ADR-0020-ai-drift-prevention.md +55 -0
- package/docs/decisions/ADR-0021-codebase-mapper-v2.md +76 -0
- package/docs/decisions/ADR-0022-lifecycle-completeness.md +63 -0
- package/docs/decisions/ADR-0023-opus-4-7-adoption.md +89 -0
- package/docs/decisions/ADR-0024-spec-b-v2-completion.md +137 -0
- package/docs/decisions/ADR-0025-behavioral-guardrails-layer.md +129 -0
- package/docs/decisions/ADR-0026-self-improvement-loop.md +171 -0
- package/docs/decisions/ADR-0027-doc-code-link-graph.md +164 -0
- package/docs/decisions/ADR-0028-unified-agents-skills-target.md +46 -0
- package/docs/decisions/ADR-0029-experiment-evaluator.md +76 -0
- package/docs/decisions/ADR-0030-planning-concurrency.md +37 -0
- package/docs/decisions/ADR-0031-focus-auto-backlog-source.md +47 -0
- package/docs/decisions/ADR-0032-squad-model.md +64 -0
- package/docs/decisions/ADR-0033-army-campaign.md +40 -0
- package/docs/decisions/ADR-0034-scheduled-campaigns.md +57 -0
- package/docs/decisions/ADR-0035-army-hud-dashboard.md +40 -0
- package/docs/decisions/ADR-0036-agentic-memory-architecture.md +106 -0
- package/docs/decisions/ADR-0037-pin-security-agents-to-opus.md +51 -0
- package/docs/decisions/ADR-0038-skill-aligned-decomposition.md +41 -0
- package/docs/decisions/ADR-0039-worktree-as-project-variant.md +37 -0
- package/docs/decisions/ADR-0040-memory-injection-defense.md +55 -0
- package/docs/decisions/ADR-0041-mcp-dual-era-bridge.md +47 -0
- package/docs/decisions/ADR-0042-design-parity-both-flows.md +45 -0
- package/docs/decisions/ADR-0043-addressable-planning-roots.md +56 -0
- package/docs/decisions/ADR-0044-bound-the-cached-context.md +104 -0
- package/docs/decisions/ADR-0045-agent-plugins-bundle.md +76 -0
- package/docs/decisions/ADR-0046-subagent-frontmatter-decisions.md +39 -0
- package/docs/decisions/ADR-0047-pan-harness.md +76 -0
- package/docs/decisions/ADR-0048-one-core-per-project.md +43 -0
- package/docs/decisions/ADR-0049-machine-verdicts-and-findings-ledger.md +116 -0
- package/docs/specs/ai_drift_prevention_featureai.md +286 -0
- package/docs/specs/codebase_mapper_v2_featureai.md +930 -0
- package/docs/specs/command_naming_optimization_featureai.md +930 -0
- package/docs/specs/comp_analysis_markdown_v1.md +149 -0
- package/docs/specs/continuous_autonomous_execution_featureai.md +304 -0
- package/docs/specs/contributors_attribution_index_featureai.md +476 -0
- package/docs/specs/copilot_cli_interaction_featureai.md +868 -0
- package/docs/specs/copilot_cli_runtime_featureai.md +647 -0
- package/docs/specs/custom-skills-assessment.md +461 -0
- package/docs/specs/deployment_audit_findings_featureai.md +243 -0
- package/docs/specs/doc_code_link_graph_featureai.md +253 -0
- package/docs/specs/documentation_alignment_cleanup_featureai.md +69 -0
- package/docs/specs/e2e_visual_testing_featureai.md +615 -0
- package/docs/specs/end_to_end_user_acceptance_testing_featureai.md +810 -0
- package/docs/specs/evidence-loop-plan.md +76 -0
- package/docs/specs/evidence_loop_featureai.md +497 -0
- package/docs/specs/focus-auto-optimize-category.md +449 -0
- package/docs/specs/focus-auto-optimize-superplan.md +188 -0
- package/docs/specs/focus_auto_runner_featureai.md +1102 -0
- package/docs/specs/focus_design_v2_featureai.md +197 -0
- package/docs/specs/focus_sync_v2_featureai.md +202 -0
- package/docs/specs/git_integration_overhaul_featureai.md +326 -0
- package/docs/specs/glm_zai_provider_featureai.md +200 -0
- package/docs/specs/googlecli_adoption_featureai.md +661 -0
- package/docs/specs/help_commands_rewrite_featureai.md +860 -0
- package/docs/specs/industry_analysis_strategic_additions_featureai.md +231 -0
- package/docs/specs/industry_comparison_and_feature_gaps_featureai.md +674 -0
- package/docs/specs/internal_cleanup_code_quality_featureai.md +486 -0
- package/docs/specs/investigations/README.md +31 -0
- package/docs/specs/investigations/digests/2026-09-28.md +47 -0
- package/docs/specs/investigations/digests/2026-10-08.md +62 -0
- package/docs/specs/investigations/digests/2026-10-10.md +61 -0
- package/docs/specs/investigations/features/debugger-red-loop-first.md +55 -0
- package/docs/specs/investigations/features/harness-cross-host-behaviour.md +93 -0
- package/docs/specs/investigations/features/optimize-apply-regression-gate.md +66 -0
- package/docs/specs/investigations/features/reviewer-smell-baseline.md +57 -0
- package/docs/specs/investigations/features/trace-failure-capture.md +69 -0
- package/docs/specs/lifecycle_completeness_featureai.md +754 -0
- package/docs/specs/local-first-ci.md +42 -0
- package/docs/specs/map_codebase_mermaid_togaf_featureai.md +805 -0
- package/docs/specs/market-delta-2026-09-superplan.md +310 -0
- package/docs/specs/market-ideas-2026-09.md +131 -0
- package/docs/specs/market-ideas-2026-10.md +428 -0
- package/docs/specs/market-ideas-ledger.md +177 -0
- package/docs/specs/memory-optimization-2026-10.md +285 -0
- package/docs/specs/multi-track-planning-roots.md +143 -0
- package/docs/specs/multi_model_routing_featureai.md +687 -0
- package/docs/specs/multi_model_routing_superplan.md +197 -0
- package/docs/specs/opus_47_existing_enhancements_featureai.md +247 -0
- package/docs/specs/opus_47_extended_features_featureai.md +363 -0
- package/docs/specs/opus_47_extended_features_v2_featureai.md +397 -0
- package/docs/specs/pan_focus_commands_featureai.md +2336 -0
- package/docs/specs/pan_zcode_mcp_bridge_featureai.md +120 -0
- package/docs/specs/product-excellence-superplan.md +315 -0
- package/docs/specs/product_excellence_program_2026_q4_featureai.md +365 -0
- package/docs/specs/production_deployment_readiness_featureai.md +433 -0
- package/docs/specs/production_readiness_ship_v1_featureai.md +698 -0
- package/docs/specs/real_test_enforcement_featureai.md +213 -0
- package/docs/specs/reality-check-2026-09-superplan.md +362 -0
- package/docs/specs/right-sized-ceremony.md +74 -0
- package/docs/specs/roadmap_and_competition_2026_q4_featureai.md +390 -0
- package/docs/specs/self_improvement_loop_featureai.md +875 -0
- package/docs/specs/skill-aligned-decomposition.md +155 -0
- package/docs/specs/standards_integration_featureai.md +833 -0
- package/docs/specs/testing-system-redesign-2026-09.md +267 -0
- package/docs/specs/user_reality_testing_system_featureai.md +788 -0
- package/docs/specs/watch-triggers.json +22 -0
- package/docs/specs/workflow-integration-redesign.md +742 -0
- package/docs/specs/workflow_state_consistency_featureai.md +183 -0
- package/hooks/dist/pan-context-monitor.js +9 -2
- package/hooks/dist/pan-cost-logger.js +5 -0
- package/hooks/dist/pan-guard.js +28 -5
- package/hooks/dist/pan-stop-guard.js +9 -9
- package/hooks/dist/pan-trace-logger.js +5 -0
- package/package.json +10 -2
- package/pan-wizard-core/bin/lib/agents-md.cjs +57 -8
- package/pan-wizard-core/bin/lib/autodev-report.cjs +180 -0
- package/pan-wizard-core/bin/lib/autodev.cjs +642 -0
- package/pan-wizard-core/bin/lib/ci.cjs +143 -0
- package/pan-wizard-core/bin/lib/commands.cjs +1 -1
- package/pan-wizard-core/bin/lib/config.cjs +41 -5
- package/pan-wizard-core/bin/lib/constants.cjs +27 -1
- package/pan-wizard-core/bin/lib/core.cjs +31 -14
- package/pan-wizard-core/bin/lib/cost.cjs +160 -4
- package/pan-wizard-core/bin/lib/cross-family.cjs +164 -0
- package/pan-wizard-core/bin/lib/debug-cases.cjs +104 -0
- package/pan-wizard-core/bin/lib/doc-lint.cjs +5 -3
- package/pan-wizard-core/bin/lib/failure-class.cjs +81 -0
- package/pan-wizard-core/bin/lib/findings.cjs +148 -9
- package/pan-wizard-core/bin/lib/focus.cjs +1 -1
- package/pan-wizard-core/bin/lib/git.cjs +1 -1
- package/pan-wizard-core/bin/lib/hosts.cjs +88 -0
- package/pan-wizard-core/bin/lib/init.cjs +7 -2
- package/pan-wizard-core/bin/lib/knowledge.cjs +2 -0
- package/pan-wizard-core/bin/lib/merge.cjs +50 -0
- package/pan-wizard-core/bin/lib/mutate.cjs +268 -0
- package/pan-wizard-core/bin/lib/next-action.cjs +112 -0
- package/pan-wizard-core/bin/lib/optimize.cjs +2 -2
- package/pan-wizard-core/bin/lib/phase-gate.cjs +15 -9
- package/pan-wizard-core/bin/lib/phase-remove.cjs +91 -11
- package/pan-wizard-core/bin/lib/phase.cjs +85 -13
- package/pan-wizard-core/bin/lib/plan-lint.cjs +310 -0
- package/pan-wizard-core/bin/lib/preview.cjs +26 -4
- package/pan-wizard-core/bin/lib/requirements-lint.cjs +130 -0
- package/pan-wizard-core/bin/lib/research-lint.cjs +268 -0
- package/pan-wizard-core/bin/lib/review-deep.cjs +14 -22
- package/pan-wizard-core/bin/lib/review-vote.cjs +105 -0
- package/pan-wizard-core/bin/lib/roadmap.cjs +1 -0
- package/pan-wizard-core/bin/lib/runner.cjs +2 -16
- package/pan-wizard-core/bin/lib/sizing.cjs +356 -0
- package/pan-wizard-core/bin/lib/squads.cjs +1 -1
- package/pan-wizard-core/bin/lib/state-write.cjs +247 -0
- package/pan-wizard-core/bin/lib/test-gate.cjs +433 -0
- package/pan-wizard-core/bin/lib/uat.cjs +149 -0
- package/pan-wizard-core/bin/lib/usage.cjs +104 -0
- package/pan-wizard-core/bin/lib/verdict.cjs +26 -3
- package/pan-wizard-core/bin/lib/verify-plan-set.cjs +30 -1
- package/pan-wizard-core/bin/lib/verify-preflight.cjs +2 -2
- package/pan-wizard-core/bin/lib/verify-red.cjs +174 -0
- package/pan-wizard-core/bin/lib/verify-retro.cjs +3 -2
- package/pan-wizard-core/bin/lib/verify-scope.cjs +143 -2
- package/pan-wizard-core/bin/lib/verify.cjs +86 -31
- package/pan-wizard-core/bin/lib/worktree.cjs +70 -0
- package/pan-wizard-core/bin/pan-tools.cjs +257 -28
- package/pan-wizard-core/learnings/universal/autonomous-loop.md +2 -2
- package/pan-wizard-core/mcp/orchestrator.cjs +4 -1
- package/pan-wizard-core/references/continuation-format.md +3 -3
- package/pan-wizard-core/references/design-methodology.md +1 -1
- package/pan-wizard-core/references/guardrails.md +1 -1
- package/pan-wizard-core/references/handoff-decisions.md +1 -1
- package/pan-wizard-core/references/model-profiles.md +2 -3
- package/pan-wizard-core/references/questioning.md +24 -0
- package/pan-wizard-core/references/ui-brand.md +4 -6
- package/pan-wizard-core/references/verdict-contract.md +2 -1
- package/pan-wizard-core/templates/config.json +10 -2
- package/pan-wizard-core/templates/debug.md +1 -0
- package/pan-wizard-core/templates/design.md +5 -4
- package/pan-wizard-core/templates/discovery.md +2 -2
- package/pan-wizard-core/templates/phase-prompt.md +7 -0
- package/pan-wizard-core/templates/requirements.md +9 -0
- package/pan-wizard-core/templates/research.md +10 -0
- package/pan-wizard-core/templates/summary-complex.md +5 -1
- package/pan-wizard-core/templates/summary-minimal.md +4 -0
- package/pan-wizard-core/templates/summary-standard.md +5 -1
- package/pan-wizard-core/templates/summary.md +8 -1
- package/pan-wizard-core/workflows/add-phase.md +4 -6
- package/pan-wizard-core/workflows/assumptions.md +19 -1
- package/pan-wizard-core/workflows/diagnose-issues.md +2 -0
- package/pan-wizard-core/workflows/discuss-phase.md +4 -2
- package/pan-wizard-core/workflows/exec-lean.md +69 -0
- package/pan-wizard-core/workflows/exec-phase.md +79 -15
- package/pan-wizard-core/workflows/execute-plan.md +3 -1
- package/{commands/pan → pan-wizard-core/workflows}/focus-doc-audit.md +13 -29
- package/{commands/pan → pan-wizard-core/workflows}/focus-drift-walking.md +3 -18
- package/{commands/pan → pan-wizard-core/workflows}/focus-exec.md +8 -23
- package/{commands/pan → pan-wizard-core/workflows}/focus-plan.md +6 -20
- package/{commands/pan → pan-wizard-core/workflows}/focus-scan.md +3 -16
- package/{commands/pan → pan-wizard-core/workflows}/focus-sync.md +7 -21
- package/pan-wizard-core/workflows/health.md +6 -0
- package/pan-wizard-core/workflows/help.md +114 -77
- package/pan-wizard-core/workflows/insert-phase.md +4 -6
- package/pan-wizard-core/workflows/learn.md +3 -3
- package/pan-wizard-core/workflows/milestone-audit.md +2 -2
- package/pan-wizard-core/workflows/milestone-new.md +38 -95
- package/pan-wizard-core/workflows/new-project.md +96 -258
- package/pan-wizard-core/workflows/optimize.md +107 -5
- package/pan-wizard-core/workflows/phase-tests.md +10 -0
- package/pan-wizard-core/workflows/plan-phase.md +106 -45
- package/pan-wizard-core/workflows/profile.md +50 -0
- package/pan-wizard-core/workflows/progress.md +10 -0
- package/pan-wizard-core/workflows/quick.md +34 -43
- package/pan-wizard-core/workflows/remove-phase.md +11 -1
- package/pan-wizard-core/workflows/research-phase.md +118 -40
- package/pan-wizard-core/workflows/resume-project.md +18 -2
- package/pan-wizard-core/workflows/settings.md +33 -19
- package/pan-wizard-core/workflows/todo-check.md +6 -2
- package/pan-wizard-core/workflows/transition.md +22 -4
- package/pan-wizard-core/workflows/uat-walkthrough.md +90 -0
- package/pan-wizard-core/workflows/update.md +1 -1
- package/pan-wizard-core/workflows/verify-phase.md +39 -46
- package/scripts/build-agent-plugin.js +35 -15
- package/scripts/build-help.cjs +106 -0
- package/scripts/build-plugin.js +10 -1
- package/scripts/deprecate-old-versions.js +70 -3
- package/scripts/generate-skills-docs.py +1 -1
- package/scripts/git-hooks/pre-push +27 -0
- package/scripts/install-git-hooks.js +7 -5
- package/scripts/install-pre-push.cjs +38 -0
- package/scripts/land.cjs +151 -0
- package/scripts/local-ci.cjs +501 -0
- package/scripts/local-schedule.cjs +277 -0
- package/scripts/mutation-probe.cjs +7 -80
- package/scripts/prompt-lint.cjs +32 -0
- package/scripts/prompt-rules.cjs +156 -0
- package/scripts/run-tests.cjs +7 -1
- package/scripts/test-quality-lint.cjs +3 -3
- package/scripts/test-surface.cjs +175 -6
- package/agents/pan-research-synthesizer.md +0 -246
- package/commands/pan/audit-deployment.md +0 -386
- package/commands/pan/dashboard.md +0 -22
- package/commands/pan/discord.md +0 -19
- package/commands/pan/mcp-bridge.md +0 -143
- package/commands/pan/milestone-cleanup.md +0 -19
- package/commands/pan/milestone-gaps.md +0 -35
- package/commands/pan/optimize.md +0 -115
- package/commands/pan/profile.md +0 -76
- package/commands/pan/research-phase.md +0 -189
- package/pan-wizard-core/bin/lib/bridge.cjs +0 -269
- package/pan-wizard-core/templates/research-project/architecture.md +0 -204
- package/pan-wizard-core/templates/research-project/features.md +0 -147
- package/pan-wizard-core/templates/research-project/pitfalls.md +0 -200
- package/pan-wizard-core/templates/research-project/stack.md +0 -120
- package/pan-wizard-core/templates/research-project/summary.md +0 -170
|
@@ -0,0 +1,119 @@
|
|
|
1
|
+
<div align="center">
|
|
2
|
+
|
|
3
|
+
<!-- Replace with the generated banner (Nano Banana prompt #1) or a 2x screenshot
|
|
4
|
+
of the README hero card in PanWizard Brand.dc.html -->
|
|
5
|
+
<img src="assets/pan-readme-hero.png" alt="PanWizard — Context engineering that makes AI coding reliable" width="820" />
|
|
6
|
+
|
|
7
|
+
# PanWizard
|
|
8
|
+
|
|
9
|
+
**Project Automation Navigator** — a lightweight workflow-automation and
|
|
10
|
+
context-engineering system for Claude Code, OpenCode, Gemini CLI, Codex and
|
|
11
|
+
Copilot CLI.
|
|
12
|
+
|
|
13
|
+
*Solves context rot — the quality degradation that happens as the model fills its
|
|
14
|
+
context window.*
|
|
15
|
+
|
|
16
|
+
[](https://www.npmjs.com/package/pan-wizard)
|
|
17
|
+
[](https://www.npmjs.com/package/pan-wizard)
|
|
18
|
+
[](LICENSE)
|
|
19
|
+
|
|
20
|
+
```bash
|
|
21
|
+
npx pan-wizard@latest
|
|
22
|
+
```
|
|
23
|
+
|
|
24
|
+
**Works on Mac, Windows, and Linux.**
|
|
25
|
+
|
|
26
|
+
</div>
|
|
27
|
+
|
|
28
|
+
---
|
|
29
|
+
|
|
30
|
+
## What is PanWizard?
|
|
31
|
+
|
|
32
|
+
PanWizard is the context-engineering layer that makes AI coding assistants
|
|
33
|
+
reliable. It breaks work into phases that fit inside the context window, hands
|
|
34
|
+
each phase to a specialist agent with exactly the context it needs, and keeps
|
|
35
|
+
durable state that survives every reset.
|
|
36
|
+
|
|
37
|
+
The complexity lives in the system, not in your workflow. Behind the scenes:
|
|
38
|
+
context engineering, XML prompt formatting, subagent orchestration, state
|
|
39
|
+
management. What you see — a few commands that just work.
|
|
40
|
+
|
|
41
|
+
```
|
|
42
|
+
planner ──▶ executor ──▶ verifier each in a fresh 200K context window
|
|
43
|
+
│ │ │
|
|
44
|
+
└────────────┴────────────┘
|
|
45
|
+
.planning/ ← persistent state: project.md · roadmap.md · state.md
|
|
46
|
+
```
|
|
47
|
+
|
|
48
|
+
## Why it works
|
|
49
|
+
|
|
50
|
+
- **Thin commands.** A handful of `/pan:` commands orchestrate everything. You
|
|
51
|
+
describe; it builds.
|
|
52
|
+
- **Specialist agents.** Planner, executor, verifier, researcher, debugger — each
|
|
53
|
+
spawned in a fresh 200K window so quality never degrades mid-task.
|
|
54
|
+
- **Durable state.** Plans, roadmap and progress live in `.planning/` and survive
|
|
55
|
+
context resets, restarts and new sessions.
|
|
56
|
+
- **Zero runtime dependencies.** Cross-platform CLI core. MIT licensed.
|
|
57
|
+
|
|
58
|
+
## Bot army — `/pan:army`
|
|
59
|
+
|
|
60
|
+
One phase is `/pan:exec-phase`. A whole goal is a **campaign**: a
|
|
61
|
+
**Mission Control** agent (`pan-conductor`) — running on the model you launched
|
|
62
|
+
with — plans the mission and delegates to specialist **squads** rather than
|
|
63
|
+
implementing anything itself.
|
|
64
|
+
|
|
65
|
+
| Tier | Squad | Role | Access |
|
|
66
|
+
| --- | --- | --- | --- |
|
|
67
|
+
| 0 | Mission Control | plan + delegate | delegation-first (instructed, not tool-enforced) |
|
|
68
|
+
| 1 | Architecture | roadmap · plan · research | read-only · parallel |
|
|
69
|
+
| 1 | Build | `pan-executor` | write · one worktree per agent |
|
|
70
|
+
| 1 | Quality | review · harden · verify | adversarial · read-only |
|
|
71
|
+
| 1 | Release | `pan-release` | always-ask · human gate |
|
|
72
|
+
| 2 | Workers | document_code · distiller | narrow, high-volume |
|
|
73
|
+
|
|
74
|
+
Six-phase loop: **Muster → Plan → Delegate → Execute → Review → Integrate → Learn** ↺
|
|
75
|
+
|
|
76
|
+
- **Concurrency:** research parallel · builds parallel but isolated in per-agent
|
|
77
|
+
worktrees · integrate serial and human-gated.
|
|
78
|
+
- **Safety harness:** nesting depth 2 · spawn + budget caps · abort kill-switch ·
|
|
79
|
+
protected main · revert never rewrite.
|
|
80
|
+
- **Scheduled campaigns:** arm a self-resuming run with a per-day budget — a human
|
|
81
|
+
still approves every merge.
|
|
82
|
+
|
|
83
|
+
```bash
|
|
84
|
+
/pan:army "ship the v1 reporting module" --source backlog --max-cycles 5
|
|
85
|
+
/pan:army --status # campaign progress
|
|
86
|
+
/pan:army --stop # graceful halt, state preserved
|
|
87
|
+
```
|
|
88
|
+
|
|
89
|
+
## Quick start
|
|
90
|
+
|
|
91
|
+
```bash
|
|
92
|
+
npx pan-wizard@latest # choose runtime + global/local
|
|
93
|
+
/pan:new-project # questions → research → requirements → roadmap
|
|
94
|
+
/pan:plan-phase 1 # research → atomic plans → verify
|
|
95
|
+
/pan:exec-phase 1 # planner → executor → verifier
|
|
96
|
+
/pan:hud # context budget · progress · cost
|
|
97
|
+
```
|
|
98
|
+
|
|
99
|
+
## Brand
|
|
100
|
+
|
|
101
|
+
| Token | Hex | Use |
|
|
102
|
+
| --- | --- | --- |
|
|
103
|
+
| Ember | `#FF5A3C` | Primary / CTAs |
|
|
104
|
+
| Conduit | `#5B4BE6` | Links, agent connectors |
|
|
105
|
+
| Verify | `#1E8E5A` | Success / verified |
|
|
106
|
+
| Butter | `#FFCE4A` | Highlights |
|
|
107
|
+
| Ink | `#211E18` | Dark surfaces, terminals |
|
|
108
|
+
| Paper | `#F3ECDD` | Light surfaces |
|
|
109
|
+
|
|
110
|
+
Type: **Gabarito** (display) + **JetBrains Mono** (code & labels).
|
|
111
|
+
Logo: node-graph mark (coral → butter + green, indigo links) + `PanWizard` wordmark.
|
|
112
|
+
|
|
113
|
+
> Banner, avatar and illustration art: generate with `nano-banana-prompts.txt`.
|
|
114
|
+
|
|
115
|
+
---
|
|
116
|
+
|
|
117
|
+
<div align="center">
|
|
118
|
+
<sub>PanWizard · Project Automation Navigator · MIT</sub>
|
|
119
|
+
</div>
|
|
@@ -0,0 +1,205 @@
|
|
|
1
|
+
# PanWizard Rebrand — Implementation Guide (Claude Code)
|
|
2
|
+
|
|
3
|
+
This ports the new brand system into the **PanWizard repo**. Each step below is a
|
|
4
|
+
copy-paste prompt for Claude Code, run from the repo root. Do them in order; each is
|
|
5
|
+
self-contained and safe to commit on its own.
|
|
6
|
+
|
|
7
|
+
> Design source of truth lives in this design project:
|
|
8
|
+
> `PanWizard Brand.dc.html` · `PanWizard UI Kit.dc.html` · `PanWizard HUD.dc.html`
|
|
9
|
+
> Generated art + SVGs are in `assets/`. Drop those files into the repo's `assets/`
|
|
10
|
+
> before running Step 2.
|
|
11
|
+
|
|
12
|
+
---
|
|
13
|
+
|
|
14
|
+
## Brand tokens (reference — paste into any prompt if drift appears)
|
|
15
|
+
|
|
16
|
+
| Token | Hex | Role |
|
|
17
|
+
| --- | --- | --- |
|
|
18
|
+
| Ember | `#FF5A3C` | Primary / brand / active |
|
|
19
|
+
| Conduit | `#5B4BE6` | Links, agent connectors, focus |
|
|
20
|
+
| Verify | `#1E8E5A` | Success / done / healthy |
|
|
21
|
+
| Butter | `#FFCE4A` | Pending / highlights |
|
|
22
|
+
| Ink | `#211E18` | Text, dark surfaces |
|
|
23
|
+
| Paper | `#F3ECDD` | Cards / light surfaces |
|
|
24
|
+
| Sand | `#E7DBC2` | Canvas / page background |
|
|
25
|
+
|
|
26
|
+
Type: **Gabarito** (display) + **JetBrains Mono** (code/labels).
|
|
27
|
+
Logo: node-graph mark — coral parent → indigo links → butter + green children.
|
|
28
|
+
|
|
29
|
+
---
|
|
30
|
+
|
|
31
|
+
## Step 1 — Drop in the new assets
|
|
32
|
+
|
|
33
|
+
Copy these files from the design project's `assets/` into the repo's `assets/`:
|
|
34
|
+
|
|
35
|
+
- `pan-mark.svg` — primary node-graph mark (transparent)
|
|
36
|
+
- `pan-logo-lockup.svg` — mark + wordmark on ink
|
|
37
|
+
- `pan-hero.png` — README / npm hero banner (1024²)
|
|
38
|
+
- `pan-avatar.png` — npm / CLI / favicon icon (rounded square)
|
|
39
|
+
- `pan-orchestration.png` — orchestration illustration
|
|
40
|
+
- `pan-docs-header.png` — soft docs-header background
|
|
41
|
+
|
|
42
|
+
```
|
|
43
|
+
Update assets/pan-logo-2000.svg and assets/pan-logo-2000-transparent.svg to the new
|
|
44
|
+
PanWizard mark: a node-graph glyph — a coral (#FF5A3C) parent node at top, two indigo
|
|
45
|
+
(#5B4BE6) connector lines fanning down to a butter (#FFCE4A) node on the lower-left and
|
|
46
|
+
a green (#1E8E5A) node on the lower-right — beside a bold rounded-sans wordmark
|
|
47
|
+
"PanWizard" ("Pan" in #FF5A3C, "Wizard" in #FBF7EE on dark / #211E18 on light). Use
|
|
48
|
+
assets/pan-logo-lockup.svg as the reference. Keep the same viewBox/dimensions the old
|
|
49
|
+
files used so every <img> reference still renders.
|
|
50
|
+
```
|
|
51
|
+
|
|
52
|
+
---
|
|
53
|
+
|
|
54
|
+
## Step 2 — README hero + brand
|
|
55
|
+
|
|
56
|
+
```
|
|
57
|
+
Replace the README.md header so the hero image is assets/pan-hero.png (the new banner).
|
|
58
|
+
Apply the PanWizard brand: keep the centered layout, swap the npm/license badge colors
|
|
59
|
+
to brand hex (version #FF5A3C, downloads #5B4BE6, license #1E8E5A, all labelColor
|
|
60
|
+
#211E18), and add a short "Brand" section near the bottom with the token table and the
|
|
61
|
+
note that banner/avatar/illustration art is generated from image-prompts.md. Use the
|
|
62
|
+
content of README-brand-draft.md from the design project as the source — do not change
|
|
63
|
+
any factual/usage copy, commands, or install instructions.
|
|
64
|
+
```
|
|
65
|
+
|
|
66
|
+
---
|
|
67
|
+
|
|
68
|
+
## Step 3 — Re-theme the HUD (light, on-brand)
|
|
69
|
+
|
|
70
|
+
File: `pan-wizard-core/bin/lib/hud.cjs`.
|
|
71
|
+
|
|
72
|
+
```
|
|
73
|
+
In pan-wizard-core/bin/lib/hud.cjs, re-theme the HUD from the current dark palette to
|
|
74
|
+
the light PanWizard brand. Make ONLY visual changes — do not touch collectHudData(),
|
|
75
|
+
the render*() data wiring, or any data shape/test.
|
|
76
|
+
|
|
77
|
+
1. Replace the :root block at the top of HUD_CSS with:
|
|
78
|
+
|
|
79
|
+
:root{
|
|
80
|
+
--bg:#E7DBC2;--panel:#FBF7EE;--panel2:#F3ECDD;--border:#E4D8C0;
|
|
81
|
+
--text:#211E18;--text2:#5C5446;--muted:#9A9180;
|
|
82
|
+
--coral:#FF5A3C;--amber:#C9A227;--cyan:#5B4BE6;--green:#1E8E5A;
|
|
83
|
+
--violet:#5B4BE6;--red:#D2431F;
|
|
84
|
+
--font:"Gabarito","Segoe UI",system-ui,sans-serif;
|
|
85
|
+
--mono:"JetBrains Mono","SFMono-Regular",Consolas,monospace;
|
|
86
|
+
}
|
|
87
|
+
|
|
88
|
+
2. Replace the .pill.* rules with light-theme equivalents:
|
|
89
|
+
|
|
90
|
+
.pill.ok{background:#E4F3EB;color:#1E8E5A;}
|
|
91
|
+
.pill.info{background:#E9E6FB;color:#5B4BE6;}
|
|
92
|
+
.pill.warn{background:#FFF3D4;color:#9A7A12;}
|
|
93
|
+
.pill.danger{background:#FBE2DB;color:#D2431F;}
|
|
94
|
+
.pill.muted{background:#EDE8DC;color:#8C8475;}
|
|
95
|
+
|
|
96
|
+
3. Add a subtle card lift: on .panel and .metric add
|
|
97
|
+
box-shadow:0 1px 3px rgba(33,30,24,0.05);
|
|
98
|
+
4. In the TIER_COLOR map, change reasoning to 'var(--coral)' so Mission Control /
|
|
99
|
+
reasoning agents read coral instead of amber.
|
|
100
|
+
5. Bump the header: in renderMission, render the node-graph mark (inline SVG from
|
|
101
|
+
assets/pan-mark.svg) to the left of the project title.
|
|
102
|
+
|
|
103
|
+
Keep all class names, structure, and the self-contained (no-network) output intact.
|
|
104
|
+
Run `node pan-wizard-core/bin/lib/... hud --stdout` mentally / `npm test` to confirm
|
|
105
|
+
renderHud still produces the same sections.
|
|
106
|
+
```
|
|
107
|
+
|
|
108
|
+
---
|
|
109
|
+
|
|
110
|
+
## Step 4 — Add the "Now building" panel (what + where)
|
|
111
|
+
|
|
112
|
+
This is the new panel that answers *what is it building* and *where in the project*.
|
|
113
|
+
|
|
114
|
+
```
|
|
115
|
+
In pan-wizard-core/bin/lib/hud.cjs, add a new "Now building" panel that renders BETWEEN
|
|
116
|
+
the mission panel and the command stack. It must show (a) WHERE — a horizontal phase
|
|
117
|
+
stepper across all roadmap phases with the current phase highlighted — and (b) WHAT —
|
|
118
|
+
the current phase name, the pan pipeline stage (research → plan → execute → verify),
|
|
119
|
+
and the in-flight build tasks (one row per active army worktree with a progress bar).
|
|
120
|
+
|
|
121
|
+
Use only data already on the object returned by collectHudData():
|
|
122
|
+
d.roadmap // [{number,name,status,plans,summaries}]
|
|
123
|
+
d.state.current_phase, d.state.current_phase_name
|
|
124
|
+
d.worktrees // [{branch,worktree}]
|
|
125
|
+
d.progress // {percent,completed,total,...}
|
|
126
|
+
|
|
127
|
+
Add this helper + renderer (mirror the existing esc()/pill()/bar() style):
|
|
128
|
+
|
|
129
|
+
function pipelineStage(p){
|
|
130
|
+
if(!p) return 'queued';
|
|
131
|
+
if(p.status==='complete') return 'verify';
|
|
132
|
+
if(p.status==='partial') return 'execute';
|
|
133
|
+
if(p.status==='researched') return 'plan';
|
|
134
|
+
if(p.status==='discussed') return 'research';
|
|
135
|
+
return 'queued';
|
|
136
|
+
}
|
|
137
|
+
|
|
138
|
+
function renderNowBuilding(d){
|
|
139
|
+
if(!d.roadmap.length) return '';
|
|
140
|
+
const curNum = d.state.current_phase
|
|
141
|
+
|| (d.roadmap.find(p=>p.status!=='complete')||{}).number;
|
|
142
|
+
const cur = d.roadmap.find(p=>String(p.number)===String(curNum)) || d.roadmap[0];
|
|
143
|
+
const stage = pipelineStage(cur);
|
|
144
|
+
const stages = ['research','plan','execute','verify'];
|
|
145
|
+
|
|
146
|
+
// WHERE — stepper
|
|
147
|
+
const steps = d.roadmap.map(p=>{
|
|
148
|
+
const done = p.status==='complete';
|
|
149
|
+
const active = String(p.number)===String(cur.number);
|
|
150
|
+
const cls = done?'ok':active?'now':'todo';
|
|
151
|
+
return `<div class="step ${cls}"><span class="sdot">${done?'✓':esc(p.number)}</span>`
|
|
152
|
+
+ `<span class="slabel">${esc((p.name||'').split(' ')[0]||p.number)}</span></div>`;
|
|
153
|
+
}).join('<span class="sline"></span>');
|
|
154
|
+
|
|
155
|
+
// WHAT — pipeline + in-flight worktrees
|
|
156
|
+
const pipe = stages.map(s=>{
|
|
157
|
+
const i = stages.indexOf(s), ci = stages.indexOf(stage);
|
|
158
|
+
const k = i<ci?'done':i===ci?'on':'off';
|
|
159
|
+
return `<span class="pstep ${k}">${esc(s)}${i<ci?' ✓':i===ci?' ●':''}</span>`;
|
|
160
|
+
}).join('<span class="pgt">›</span>');
|
|
161
|
+
|
|
162
|
+
const tasks = d.worktrees.length
|
|
163
|
+
? d.worktrees.map(t=>`<div class="task"><span class="tname">${esc(t.branch.replace(/^army\//,''))}</span>`
|
|
164
|
+
+ `<span class="amono dim">${esc(t.worktree)}</span></div>`).join('')
|
|
165
|
+
: `<div class="task dim amono">${cur.plans||0} plan(s) · ${cur.summaries||0} done</div>`;
|
|
166
|
+
|
|
167
|
+
return `
|
|
168
|
+
<section class="panel nowbuilding">
|
|
169
|
+
<div class="ph">now building — phase ${esc(cur.number)} of ${esc(d.progress.total)}</div>
|
|
170
|
+
<div class="stepper">${steps}</div>
|
|
171
|
+
<div class="nbcard">
|
|
172
|
+
<div class="nbhead"><div class="nbtitle">Phase ${esc(cur.number)} — ${esc(cur.name||'')}</div>
|
|
173
|
+
<div class="pipeline">${pipe}</div></div>
|
|
174
|
+
<div class="tasks">${tasks}</div>
|
|
175
|
+
</div>
|
|
176
|
+
</section>`;
|
|
177
|
+
}
|
|
178
|
+
|
|
179
|
+
Wire it into renderHud()'s body array, right after renderMission(d). Add matching CSS to
|
|
180
|
+
HUD_CSS using existing tokens (.stepper as a flex row; .step.now .sdot uses --coral with
|
|
181
|
+
a soft ring; .step.ok .sdot uses --green; .pstep.on uses --coral). Then add
|
|
182
|
+
'now-building' to the sections list in cmdHud() so the JSON result reports it.
|
|
183
|
+
Add a unit test in tests/hud.test.cjs asserting renderHud output contains "now building"
|
|
184
|
+
when a roadmap exists.
|
|
185
|
+
```
|
|
186
|
+
|
|
187
|
+
---
|
|
188
|
+
|
|
189
|
+
## Step 5 — Verify
|
|
190
|
+
|
|
191
|
+
```
|
|
192
|
+
Run npm test and npm run test:scenarios. Then generate a sample HUD with
|
|
193
|
+
`node bin/install.js` env or the pan-tools dispatcher: `pan-tools hud --out /tmp/hud.html`
|
|
194
|
+
and open it to confirm the light theme + the "now building" panel render correctly with
|
|
195
|
+
real project data. Fix any failing hud tests caused by the new section.
|
|
196
|
+
```
|
|
197
|
+
|
|
198
|
+
---
|
|
199
|
+
|
|
200
|
+
## Notes
|
|
201
|
+
- The HUD stays a single self-contained file — no external CSS/JS/fonts fetched. Gabarito
|
|
202
|
+
/ JetBrains Mono are named with system fallbacks; if you want them embedded, base64 a
|
|
203
|
+
woff2 into an `@font-face` inside HUD_CSS (optional).
|
|
204
|
+
- Nothing here changes planning/state semantics — the HUD remains a pure view.
|
|
205
|
+
- Re-generate any new art with the prompts in `image-prompts.md`.
|
|
@@ -0,0 +1,119 @@
|
|
|
1
|
+
<div align="center">
|
|
2
|
+
|
|
3
|
+
<!-- Replace with the generated banner (Nano Banana prompt #1) or a 2x screenshot
|
|
4
|
+
of the README hero card in PanWizard Brand.dc.html -->
|
|
5
|
+
<img src="assets/pan-readme-hero.png" alt="PanWizard — Context engineering that makes AI coding reliable" width="820" />
|
|
6
|
+
|
|
7
|
+
# PanWizard
|
|
8
|
+
|
|
9
|
+
**Project Automation Navigator** — a lightweight workflow-automation and
|
|
10
|
+
context-engineering system for Claude Code, OpenCode, Gemini CLI, Codex and
|
|
11
|
+
Copilot CLI.
|
|
12
|
+
|
|
13
|
+
*Solves context rot — the quality degradation that happens as the model fills its
|
|
14
|
+
context window.*
|
|
15
|
+
|
|
16
|
+
[](https://www.npmjs.com/package/pan-wizard)
|
|
17
|
+
[](https://www.npmjs.com/package/pan-wizard)
|
|
18
|
+
[](LICENSE)
|
|
19
|
+
|
|
20
|
+
```bash
|
|
21
|
+
npx pan-wizard@latest
|
|
22
|
+
```
|
|
23
|
+
|
|
24
|
+
**Works on Mac, Windows, and Linux.**
|
|
25
|
+
|
|
26
|
+
</div>
|
|
27
|
+
|
|
28
|
+
---
|
|
29
|
+
|
|
30
|
+
## What is PanWizard?
|
|
31
|
+
|
|
32
|
+
PanWizard is the context-engineering layer that makes AI coding assistants
|
|
33
|
+
reliable. It breaks work into phases that fit inside the context window, hands
|
|
34
|
+
each phase to a specialist agent with exactly the context it needs, and keeps
|
|
35
|
+
durable state that survives every reset.
|
|
36
|
+
|
|
37
|
+
The complexity lives in the system, not in your workflow. Behind the scenes:
|
|
38
|
+
context engineering, XML prompt formatting, subagent orchestration, state
|
|
39
|
+
management. What you see — a few commands that just work.
|
|
40
|
+
|
|
41
|
+
```
|
|
42
|
+
planner ──▶ executor ──▶ verifier each in a fresh 200K context window
|
|
43
|
+
│ │ │
|
|
44
|
+
└────────────┴────────────┘
|
|
45
|
+
.planning/ ← persistent state: project.md · roadmap.md · state.md
|
|
46
|
+
```
|
|
47
|
+
|
|
48
|
+
## Why it works
|
|
49
|
+
|
|
50
|
+
- **Thin commands.** A handful of `/pan:` commands orchestrate everything. You
|
|
51
|
+
describe; it builds.
|
|
52
|
+
- **Specialist agents.** Planner, executor, verifier, researcher, debugger — each
|
|
53
|
+
spawned in a fresh 200K window so quality never degrades mid-task.
|
|
54
|
+
- **Durable state.** Plans, roadmap and progress live in `.planning/` and survive
|
|
55
|
+
context resets, restarts and new sessions.
|
|
56
|
+
- **Zero runtime dependencies.** Cross-platform CLI core. MIT licensed.
|
|
57
|
+
|
|
58
|
+
## Bot army — `/pan:army`
|
|
59
|
+
|
|
60
|
+
One phase is `/pan:exec-phase`. A whole goal is a **campaign**: a
|
|
61
|
+
**Mission Control** agent (`pan-conductor`) — running on the model you launched
|
|
62
|
+
with — plans the mission and delegates to specialist **squads** rather than
|
|
63
|
+
implementing anything itself.
|
|
64
|
+
|
|
65
|
+
| Tier | Squad | Role | Access |
|
|
66
|
+
| --- | --- | --- | --- |
|
|
67
|
+
| 0 | Mission Control | plan + delegate | delegation-first (instructed, not tool-enforced) |
|
|
68
|
+
| 1 | Architecture | roadmap · plan · research | read-only · parallel |
|
|
69
|
+
| 1 | Build | `pan-executor` | write · one worktree per agent |
|
|
70
|
+
| 1 | Quality | review · harden · verify | adversarial · read-only |
|
|
71
|
+
| 1 | Release | `pan-release` | always-ask · human gate |
|
|
72
|
+
| 2 | Workers | document_code · distiller | narrow, high-volume |
|
|
73
|
+
|
|
74
|
+
Six-phase loop: **Muster → Plan → Delegate → Execute → Review → Integrate → Learn** ↺
|
|
75
|
+
|
|
76
|
+
- **Concurrency:** research parallel · builds parallel but isolated in per-agent
|
|
77
|
+
worktrees · integrate serial and human-gated.
|
|
78
|
+
- **Safety harness:** nesting depth 2 · spawn + budget caps · abort kill-switch ·
|
|
79
|
+
protected main · revert never rewrite.
|
|
80
|
+
- **Scheduled campaigns:** arm a self-resuming run with a per-day budget — a human
|
|
81
|
+
still approves every merge.
|
|
82
|
+
|
|
83
|
+
```bash
|
|
84
|
+
/pan:army "ship the v1 reporting module" --source backlog --max-cycles 5
|
|
85
|
+
/pan:army --status # campaign progress
|
|
86
|
+
/pan:army --stop # graceful halt, state preserved
|
|
87
|
+
```
|
|
88
|
+
|
|
89
|
+
## Quick start
|
|
90
|
+
|
|
91
|
+
```bash
|
|
92
|
+
npx pan-wizard@latest # choose runtime + global/local
|
|
93
|
+
/pan:new-project # questions → research → requirements → roadmap
|
|
94
|
+
/pan:plan-phase 1 # research → atomic plans → verify
|
|
95
|
+
/pan:exec-phase 1 # planner → executor → verifier
|
|
96
|
+
/pan:hud # context budget · progress · cost
|
|
97
|
+
```
|
|
98
|
+
|
|
99
|
+
## Brand
|
|
100
|
+
|
|
101
|
+
| Token | Hex | Use |
|
|
102
|
+
| --- | --- | --- |
|
|
103
|
+
| Ember | `#FF5A3C` | Primary / CTAs |
|
|
104
|
+
| Conduit | `#5B4BE6` | Links, agent connectors |
|
|
105
|
+
| Verify | `#1E8E5A` | Success / verified |
|
|
106
|
+
| Butter | `#FFCE4A` | Highlights |
|
|
107
|
+
| Ink | `#211E18` | Dark surfaces, terminals |
|
|
108
|
+
| Paper | `#F3ECDD` | Light surfaces |
|
|
109
|
+
|
|
110
|
+
Type: **Gabarito** (display) + **JetBrains Mono** (code & labels).
|
|
111
|
+
Logo: node-graph mark (coral → butter + green, indigo links) + `PanWizard` wordmark.
|
|
112
|
+
|
|
113
|
+
> Banner, avatar and illustration art: generate with `nano-banana-prompts.txt`.
|
|
114
|
+
|
|
115
|
+
---
|
|
116
|
+
|
|
117
|
+
<div align="center">
|
|
118
|
+
<sub>PanWizard · Project Automation Navigator · MIT</sub>
|
|
119
|
+
</div>
|
|
@@ -0,0 +1,46 @@
|
|
|
1
|
+
# PanWizard — Image Prompts for Nano Banana (Gemini)
|
|
2
|
+
|
|
3
|
+
Hand any of these to Nano Banana. Every prompt locks the brand palette so output stays on-system.
|
|
4
|
+
|
|
5
|
+
## Brand constraints (paste with any prompt if drift appears)
|
|
6
|
+
- **Palette only:** Ember coral `#FF5A3C` · Conduit indigo `#5B4BE6` · Verify green `#1E8E5A` · Butter yellow `#FFCE4A` · Ink `#211E18` · Paper cream `#F3ECDD` / `#FBF7EE`
|
|
7
|
+
- **Logo motif:** a node-graph — one coral dot branching *down* to a butter dot and a green dot, joined by indigo connector lines (planner → executor → verifier)
|
|
8
|
+
- **Type vibe:** bold rounded geometric sans (Gabarito) + JetBrains Mono
|
|
9
|
+
- **Style:** flat vector, rounded corners (8 / 16 / 32px), soft shadows, soft grain, generous negative space
|
|
10
|
+
- **Avoid:** neon, heavy gradients, 3D (unless asked), stock-photo people (unless asked), busy backgrounds
|
|
11
|
+
|
|
12
|
+
---
|
|
13
|
+
|
|
14
|
+
## 1 · GitHub social preview / npm hero — `1280 × 640`
|
|
15
|
+
> Wide developer-brand banner on a deep charcoal-brown background `#211E18`. A clean geometric node-graph logo: one coral `#FF5A3C` dot branching down to a butter `#FFCE4A` dot and a green `#1E8E5A` dot, connected by indigo `#5B4BE6` lines. To its right, a bold rounded sans wordmark "PanWizard" with "Pan" in coral and "Wizard" in cream. Subhead: "Context engineering that makes AI coding reliable." A small cream pill reading `npx pan-wizard@latest` in monospace. Flat vector, soft grain, generous negative space, no photo, no 3D. Friendly, premium dev-tool aesthetic.
|
|
16
|
+
|
|
17
|
+
## 2 · Hero illustration — orchestration
|
|
18
|
+
> Isometric flat illustration of three friendly rounded robot agents labelled planner, executor and verifier, passing a small glowing token along a connected track. Warm cream `#F3ECDD` background. Strict palette: coral `#FF5A3C`, indigo `#5B4BE6`, green `#1E8E5A`, butter `#FFCE4A`, ink `#211E18`. Rounded shapes, 16–32px corner radii, soft shadows, no heavy gradients, no neon. Calm, approachable, editorial vector style with lots of breathing room.
|
|
19
|
+
|
|
20
|
+
## 3 · App / CLI avatar — square `1024 × 1024`
|
|
21
|
+
> Square app icon, rounded-square tile filled coral `#FF5A3C`, a centered minimal node-graph glyph in cream `#FBF7EE` (one dot branching to two dots with thin connector lines), subtle inner shadow, flat and crisp, no text. iOS-style icon grid with safe-area padding.
|
|
22
|
+
|
|
23
|
+
## 4 · Docs section header — wide, soft background
|
|
24
|
+
> Abstract soft background for documentation headers: a warm cream `#F3ECDD` field with a few large, out-of-focus rounded shapes in coral, indigo, green and butter drifting near the edges. Large empty center for a text overlay. Flat, gentle grain, no objects, no people.
|
|
25
|
+
|
|
26
|
+
## 5 · Feature spot illustrations (set of 3, square)
|
|
27
|
+
> Three matching minimal spot illustrations on cream `#F3ECDD`, one per feature, same line-weight and palette (`#FF5A3C` `#5B4BE6` `#1E8E5A` `#FFCE4A` `#211E18`):
|
|
28
|
+
> 1. **Thin commands** — a terminal window with a single glowing prompt line.
|
|
29
|
+
> 2. **Specialist agents** — three rounded robot heads connected by indigo lines.
|
|
30
|
+
> 3. **Durable state** — a stack of document cards with a small save/checkpoint badge.
|
|
31
|
+
> Flat vector, rounded corners, soft shadow, generous negative space, no text.
|
|
32
|
+
|
|
33
|
+
## 6 · OG / Twitter card — `1200 × 630`
|
|
34
|
+
> Social share card, ink `#211E18` background, node-graph logo top-left, big rounded-sans headline "Stop fighting context rot." in cream, a coral `#FF5A3C` accent underline, and a monospace `npx pan-wizard@latest` chip bottom-left. Flat, grainy, minimal.
|
|
35
|
+
|
|
36
|
+
## 7 · (Optional) friendly developer photo — brand-tinted
|
|
37
|
+
> Natural candid photo of a relaxed developer at a desk, warm daylight, shallow depth of field. Color-grade toward the brand: warm cream tones with a coral `#FF5A3C` accent object in frame (mug or sticky note) and an indigo `#5B4BE6` element. Soft, friendly, editorial. Leave clean negative space on one side for text overlay.
|
|
38
|
+
|
|
39
|
+
---
|
|
40
|
+
|
|
41
|
+
### Where each image goes
|
|
42
|
+
- **#1** → repo `assets/pan-readme-hero.png`, referenced at the top of `README.md`
|
|
43
|
+
- **#3** → npm/GitHub avatar, CLI icon, favicon
|
|
44
|
+
- **#6** → `og:image` / Twitter card meta
|
|
45
|
+
- **#2, #4, #5** → landing page + docs site
|
|
46
|
+
- **#7** → optional landing-page section (only if you want a human touch)
|
|
@@ -0,0 +1,46 @@
|
|
|
1
|
+
# PanWizard — Image Prompts for Nano Banana (Gemini)
|
|
2
|
+
|
|
3
|
+
Hand any of these to Nano Banana. Every prompt locks the brand palette so output stays on-system.
|
|
4
|
+
|
|
5
|
+
## Brand constraints (paste with any prompt if drift appears)
|
|
6
|
+
- **Palette only:** Ember coral `#FF5A3C` · Conduit indigo `#5B4BE6` · Verify green `#1E8E5A` · Butter yellow `#FFCE4A` · Ink `#211E18` · Paper cream `#F3ECDD` / `#FBF7EE`
|
|
7
|
+
- **Logo motif:** a node-graph — one coral dot branching *down* to a butter dot and a green dot, joined by indigo connector lines (planner → executor → verifier)
|
|
8
|
+
- **Type vibe:** bold rounded geometric sans (Gabarito) + JetBrains Mono
|
|
9
|
+
- **Style:** flat vector, rounded corners (8 / 16 / 32px), soft shadows, soft grain, generous negative space
|
|
10
|
+
- **Avoid:** neon, heavy gradients, 3D (unless asked), stock-photo people (unless asked), busy backgrounds
|
|
11
|
+
|
|
12
|
+
---
|
|
13
|
+
|
|
14
|
+
## 1 · GitHub social preview / npm hero — `1280 × 640`
|
|
15
|
+
> Wide developer-brand banner on a deep charcoal-brown background `#211E18`. A clean geometric node-graph logo: one coral `#FF5A3C` dot branching down to a butter `#FFCE4A` dot and a green `#1E8E5A` dot, connected by indigo `#5B4BE6` lines. To its right, a bold rounded sans wordmark "PanWizard" with "Pan" in coral and "Wizard" in cream. Subhead: "Context engineering that makes AI coding reliable." A small cream pill reading `npx pan-wizard@latest` in monospace. Flat vector, soft grain, generous negative space, no photo, no 3D. Friendly, premium dev-tool aesthetic.
|
|
16
|
+
|
|
17
|
+
## 2 · Hero illustration — orchestration
|
|
18
|
+
> Isometric flat illustration of three friendly rounded robot agents labelled planner, executor and verifier, passing a small glowing token along a connected track. Warm cream `#F3ECDD` background. Strict palette: coral `#FF5A3C`, indigo `#5B4BE6`, green `#1E8E5A`, butter `#FFCE4A`, ink `#211E18`. Rounded shapes, 16–32px corner radii, soft shadows, no heavy gradients, no neon. Calm, approachable, editorial vector style with lots of breathing room.
|
|
19
|
+
|
|
20
|
+
## 3 · App / CLI avatar — square `1024 × 1024`
|
|
21
|
+
> Square app icon, rounded-square tile filled coral `#FF5A3C`, a centered minimal node-graph glyph in cream `#FBF7EE` (one dot branching to two dots with thin connector lines), subtle inner shadow, flat and crisp, no text. iOS-style icon grid with safe-area padding.
|
|
22
|
+
|
|
23
|
+
## 4 · Docs section header — wide, soft background
|
|
24
|
+
> Abstract soft background for documentation headers: a warm cream `#F3ECDD` field with a few large, out-of-focus rounded shapes in coral, indigo, green and butter drifting near the edges. Large empty center for a text overlay. Flat, gentle grain, no objects, no people.
|
|
25
|
+
|
|
26
|
+
## 5 · Feature spot illustrations (set of 3, square)
|
|
27
|
+
> Three matching minimal spot illustrations on cream `#F3ECDD`, one per feature, same line-weight and palette (`#FF5A3C` `#5B4BE6` `#1E8E5A` `#FFCE4A` `#211E18`):
|
|
28
|
+
> 1. **Thin commands** — a terminal window with a single glowing prompt line.
|
|
29
|
+
> 2. **Specialist agents** — three rounded robot heads connected by indigo lines.
|
|
30
|
+
> 3. **Durable state** — a stack of document cards with a small save/checkpoint badge.
|
|
31
|
+
> Flat vector, rounded corners, soft shadow, generous negative space, no text.
|
|
32
|
+
|
|
33
|
+
## 6 · OG / Twitter card — `1200 × 630`
|
|
34
|
+
> Social share card, ink `#211E18` background, node-graph logo top-left, big rounded-sans headline "Stop fighting context rot." in cream, a coral `#FF5A3C` accent underline, and a monospace `npx pan-wizard@latest` chip bottom-left. Flat, grainy, minimal.
|
|
35
|
+
|
|
36
|
+
## 7 · (Optional) friendly developer photo — brand-tinted
|
|
37
|
+
> Natural candid photo of a relaxed developer at a desk, warm daylight, shallow depth of field. Color-grade toward the brand: warm cream tones with a coral `#FF5A3C` accent object in frame (mug or sticky note) and an indigo `#5B4BE6` element. Soft, friendly, editorial. Leave clean negative space on one side for text overlay.
|
|
38
|
+
|
|
39
|
+
---
|
|
40
|
+
|
|
41
|
+
### Where each image goes
|
|
42
|
+
- **#1** → repo `assets/pan-readme-hero.png`, referenced at the top of `README.md`
|
|
43
|
+
- **#3** → npm/GitHub avatar, CLI icon, favicon
|
|
44
|
+
- **#6** → `og:image` / Twitter card meta
|
|
45
|
+
- **#2, #4, #5** → landing page + docs site
|
|
46
|
+
- **#7** → optional landing-page section (only if you want a human touch)
|
|
@@ -0,0 +1,60 @@
|
|
|
1
|
+
PANWIZARD — NANO BANANA (GEMINI) IMAGE PROMPTS
|
|
2
|
+
================================================
|
|
3
|
+
Brand palette (keep locked in every prompt):
|
|
4
|
+
Ember coral #FF5A3C (primary)
|
|
5
|
+
Conduit indigo #5B4BE6 (links / secondary)
|
|
6
|
+
Verify green #1E8E5A
|
|
7
|
+
Butter yellow #FFCE4A
|
|
8
|
+
Ink #211E18 (dark backgrounds)
|
|
9
|
+
Paper cream #F3ECDD / #FBF7EE
|
|
10
|
+
Logo motif: a node-graph — one coral dot branching DOWN to a yellow dot and a
|
|
11
|
+
green dot, joined by indigo connector lines (planner -> executor -> verifier).
|
|
12
|
+
Type vibe: bold rounded geometric sans (Gabarito) + JetBrains Mono.
|
|
13
|
+
Style rules: flat vector, rounded corners (8/16/32px), soft shadows, soft grain,
|
|
14
|
+
generous negative space. NO neon, NO heavy gradients, NO 3D unless asked,
|
|
15
|
+
NO stock-photo people unless asked.
|
|
16
|
+
|
|
17
|
+
|
|
18
|
+
1) GITHUB SOCIAL PREVIEW / NPM HERO (1280 x 640)
|
|
19
|
+
-------------------------------------------------
|
|
20
|
+
Wide developer-brand banner on a deep charcoal-brown background #211E18. A clean
|
|
21
|
+
geometric node-graph logo: one coral #FF5A3C dot branching down to a butter
|
|
22
|
+
#FFCE4A dot and a green #1E8E5A dot, connected by indigo #5B4BE6 lines. To its
|
|
23
|
+
right, a bold rounded sans wordmark "PanWizard" with "Pan" in coral and "Wizard"
|
|
24
|
+
in cream. Subhead "Context engineering that makes AI coding reliable." A small
|
|
25
|
+
cream pill reading "npx pan-wizard@latest" in monospace. Flat vector, soft grain,
|
|
26
|
+
generous negative space, no photo, no 3D. Friendly, premium dev-tool aesthetic.
|
|
27
|
+
|
|
28
|
+
|
|
29
|
+
2) HERO ILLUSTRATION — ORCHESTRATION
|
|
30
|
+
------------------------------------
|
|
31
|
+
Isometric flat illustration of three friendly rounded robot agents labelled
|
|
32
|
+
planner, executor and verifier, passing a small glowing token along a connected
|
|
33
|
+
track. Warm cream #F3ECDD background. Strict palette: coral #FF5A3C, indigo
|
|
34
|
+
#5B4BE6, green #1E8E5A, butter #FFCE4A, ink #211E18. Rounded shapes, 16-32px
|
|
35
|
+
corner radii, soft shadows, no heavy gradients, no neon. Calm, approachable,
|
|
36
|
+
editorial vector style. Lots of breathing room.
|
|
37
|
+
|
|
38
|
+
|
|
39
|
+
3) APP / CLI AVATAR (square, 1024 x 1024)
|
|
40
|
+
------------------------------------------
|
|
41
|
+
Square app icon, rounded-square tile filled coral #FF5A3C, a centered minimal
|
|
42
|
+
node-graph glyph in cream #FBF7EE (one dot branching to two dots with thin
|
|
43
|
+
connector lines), subtle inner shadow, flat and crisp, no text. iOS-style icon
|
|
44
|
+
grid with safe-area padding.
|
|
45
|
+
|
|
46
|
+
|
|
47
|
+
4) DOCS SECTION HEADER (wide, soft background)
|
|
48
|
+
-----------------------------------------------
|
|
49
|
+
Abstract soft background for documentation headers: a warm cream #F3ECDD field
|
|
50
|
+
with a few large, out-of-focus rounded shapes in coral, indigo, green and butter
|
|
51
|
+
drifting near the edges. Large empty center for a text overlay. Flat, gentle
|
|
52
|
+
grain, no objects, no people.
|
|
53
|
+
|
|
54
|
+
|
|
55
|
+
5) OPTIONAL — FRIENDLY DEVELOPER PHOTO (brand-tinted)
|
|
56
|
+
-----------------------------------------------------
|
|
57
|
+
Natural candid photo of a relaxed developer at a desk, warm daylight, shallow
|
|
58
|
+
depth of field. Color-grade toward the brand: warm cream tones, with a coral
|
|
59
|
+
#FF5A3C accent object in frame (mug, sticky note) and an indigo #5B4BE6 element.
|
|
60
|
+
Soft, friendly, editorial. Leave clean negative space on one side for text.
|
|
@@ -0,0 +1,13 @@
|
|
|
1
|
+
# Context Window Monitor
|
|
2
|
+
|
|
3
|
+
> **Note:** This page is a short summary. [HOOKS.md](HOOKS.md#pan-context-monitorjs) is the full, maintained description of the context monitor and of every other built-in hook.
|
|
4
|
+
|
|
5
|
+
`pan-context-monitor.js` is a `PostToolUse` hook. It tells the agent when the host will compact its context soon, so the agent can make sure `.planning/state.md` records where it is first.
|
|
6
|
+
|
|
7
|
+
- **What it measures:** the room left before the host compacts. The host compacts about 33K tokens short of its auto-compact window: `CLAUDE_CODE_AUTO_COMPACT_WINDOW`, then `autoCompactWindow` in settings, else the model window. `CLAUDE_AUTOCOMPACT_PCT_OVERRIDE` lowers that point when its share of the window is lower. With auto-compact off (`DISABLE_AUTO_COMPACT=1`, or `autoCompactEnabled: false` in settings) it measures the room left in the model window instead.
|
|
8
|
+
- **Where it reads it:** PAN's status line writes a bridge file, `<os-tmpdir>/pan-hooks-{uid}/claude-ctx-{session_id}.json`, and a main-thread call reads it while it is fresh. Without one (headless `claude -p`, or a status line that is not PAN's), the hook reads the session transcript the payload names. A call inside a subagent is measured on the subagent's own transcript.
|
|
9
|
+
- **What it says:** at 35% or less of that room left, a note asking for a checkpoint in `.planning/state.md` (or, in a project without one, wherever it records its progress) at the next natural stopping point; at 25% or less, a checkpoint before the next step (`/pan:pause` writes it). The note carries no countdown and asks the agent to keep working at full quality, because PAN restores the planning state after the compaction.
|
|
10
|
+
- **Where it runs:** Claude Code only. PAN registers it on no other runtime: Codex has no status-line bridge and its transcript is not Claude-shaped, Copilot's payload is camelCase, Gemini CLI exposes no context usage, and OpenCode runs no hooks (see [HOOKS.md](HOOKS.md#hook-runtime-support)).
|
|
11
|
+
- **Safety:** it fails open. Any error, a missing transcript or an unknown model window means no note, and it never blocks a tool call.
|
|
12
|
+
|
|
13
|
+
The installer registers it; [HOOKS.md](HOOKS.md#installation) shows the `settings.json` entries it writes.
|