feat: prepare ARR for controlled public deployment

.project-docs/05-agent-entry/context-checklist.md

@@ -0,0 +1,12 @@
+# Context Checklist
+
+Before planning, confirm:
+
+- I know what this project is and what it is not.
+- I know the current focus and recently completed work.
+- I checked active decisions that constrain this task.
+- I checked the current architecture overview.
+- I identified task-specific docs that need deeper reading.
+- I checked evidence, reflection, or commitment indexes when the task touches investigations, repeated workflow lessons, or follow-ups.
+- I can name unknown, stale, or conflicting information.
+- I know whether the task requires human confirmation before changing durable docs.

.project-docs/05-agent-entry/memory-index.md

@@ -0,0 +1,25 @@
+# Memory Index
+
+Use this as the high-density entry point before reading slower documents.
+
+## Startup Set
+
+- Project identity: `.project-docs/00-brief/project-positioning.md`
+- Current state: `.project-docs/30-worklog/current-state.md`
+- Decision list: `.project-docs/10-decisions/decision-index.md`
+- System shape: `.project-docs/20-architecture/system-overview.md`
+
+## Recall Pointers
+
+- Confirmed report workflow and price formula: `.project-docs/10-decisions/ADR-001-automatic-monthly-trigger-and-total-price-formula.md`
+- Business rules and exact field semantics: `.project-docs/40-domain/business-rules.md`
+- Evidence-heavy bugs, experiments, investigations: `.project-docs/50-evidence/evidence-index.md`
+- Current upload E2E blocker: `.project-docs/50-evidence/topics/2026-07-29-live-synthetic-xml-vertical-slice.md`
+- Controlled public deployment snapshot: `.project-docs/50-evidence/topics/2026-07-29-public-deployment-repository.md`
+- Workflow lessons and repeated agent mistakes: `.project-docs/60-reflection/reflection-index.md`
+- Pending promises, loops, timed follow-ups: `.project-docs/80-commitments/commitments.md`
+- Stale or conflicting docs: `.project-docs/90-maintenance/stale-items.md`
+
+## Loading Rule
+
+Keep this file short. Add pointers here, not full history.

.project-docs/05-agent-entry/planning-gate.md

@@ -0,0 +1,44 @@
+# Planning Gate
+
+A coding agent must complete this gate before writing an implementation plan.
+
+## Required Output
+
+```md
+## Project Context Loaded
+
+Read:
+- {file path}
+
+Relevant understanding:
+- Project goal:
+- Current focus:
+- Active constraints:
+- Decisions affecting this task:
+- Evidence, reflections, or commitments affecting this task:
+- Files or modules likely involved:
+- Unknowns, stale docs, or conflicts:
+
+Gate result:
+- Passed or Blocked
+```
+
+## Pass Criteria
+
+The gate passes only when:
+
+- required documents were read
+- `memory-index.md` was checked when present
+- task-relevant decisions were checked
+- relevant evidence, reflection, and commitment indexes were checked when applicable
+- stale or conflicting docs were called out
+- the plan respects project positioning and constraints
+
+## Block Criteria
+
+Block planning when:
+
+- required documents are missing and cannot be inferred
+- current state conflicts with the user request
+- an existing decision appears to be violated
+- the task changes project positioning or architecture without human confirmation

.project-docs/05-agent-entry/read-before-coding.md

@@ -0,0 +1,7 @@
+# Read Before Coding
+
+Before editing code, verify that the implementation plan already passed the planning gate.
+
+If the plan is stale, missing, or based on incomplete docs, return to `.project-docs/05-agent-entry/read-before-planning.md`.
+
+Before editing files, read the project docs and source files directly related to the target modules. If implementation reveals stale project docs, record the discrepancy in `.project-docs/90-maintenance/stale-items.md` or update the relevant docs after the task completes.

.project-docs/05-agent-entry/read-before-planning.md

@@ -0,0 +1,22 @@
+# Read Before Planning
+
+Before writing any coding plan, read these files in order:
+
+1. `.project-docs/05-agent-entry/memory-index.md`
+2. `.project-docs/00-brief/project-positioning.md`
+3. `.project-docs/30-worklog/current-state.md`
+4. `.project-docs/10-decisions/decision-index.md`
+5. `.project-docs/20-architecture/system-overview.md`
+
+Then read additional files when relevant:
+
+- Architecture or refactor task: `.project-docs/20-architecture/module-map.md` and `.project-docs/20-architecture/data-flow.md`
+- Product or behavior task: `.project-docs/40-domain/business-rules.md` and `.project-docs/00-brief/success-criteria.md`
+- Ambiguous terms: `.project-docs/40-domain/glossary.md`
+- Decision-sensitive task: referenced ADR files in `.project-docs/10-decisions/`
+- Evidence-heavy bug, investigation, or experiment: `.project-docs/50-evidence/evidence-index.md`
+- Repeated workflow issue, skipped gate, or skill/script candidate: `.project-docs/60-reflection/reflection-index.md`
+- Follow-up, loop, timed check, or restart-point task: `.project-docs/80-commitments/commitments.md`
+- Suspicious context: `.project-docs/90-maintenance/stale-items.md`
+
+Do not write a plan until the planning gate passes.