diff --git a/docs/evidence/ARCHITECTURE_OVERVIEW_CONTRIBUTORS.md b/docs/evidence/ARCHITECTURE_OVERVIEW_CONTRIBUTORS.md
new file mode 100644
index 00000000..7076a4f7
--- /dev/null
+++ b/docs/evidence/ARCHITECTURE_OVERVIEW_CONTRIBUTORS.md
@@ -0,0 +1,27 @@
+# Architecture Overview for Contributors
+
+A quick tour of the codebase structure.
+
+```
+src/openamp_foundry/     # Pipeline code
+  cli/                   # CLI commands (rank, validate, bench, etc.)
+  pipeline.py            # Main scoring pipeline
+  scoring/               # Individual scorers (activity, safety, etc.)
+  features/              # Physicochemical feature extraction
+  simulation/            # Virtual assay modules (membrane, structure)
+  calibration/           # Recalibration engine and gate
+  evidence/              # Evidence certificate generation
+  benchmark/             # Benchmark infrastructure
+tests/                   # Test suite
+  benchmarks/            # Benchmark regression tests
+  calibration/           # Calibration tests
+  lab/                   # Lab handoff tests
+configs/                 # Configuration files (pipeline.yaml, etc.)
+schemas/                 # JSON schemas for artifacts
+scripts/                 # Utility scripts
+docs/                    # Documentation
+```
+
+## Entry Points
+- `src/openamp_foundry/cli/main.py` — CLI entry points.
+- `src/openamp_foundry/pipeline.py` — `run_ranking_pipeline()` is the main function.
diff --git a/docs/evidence/CATALOG_AGENT_ISSUES.md b/docs/evidence/CATALOG_AGENT_ISSUES.md
new file mode 100644
index 00000000..655d1074
--- /dev/null
+++ b/docs/evidence/CATALOG_AGENT_ISSUES.md
@@ -0,0 +1,21 @@
+# Agent-Safe Issue Examples Catalog
+
+Example issues that are safe for AI agents to work on.
+
+## Docs
+- Fix broken links in a document
+- Add missing concept card for an artifact type
+- Update QUICKSTART.md for a CLI change
+- Add a test for a documentation file
+- Create a runbook for a contributor workflow
+
+## Tests
+- Add missing-file error test for a CLI command
+- Add metric polarity test
+- Add schema validation test
+- Add CLI exit code test
+
+## Scripts
+- Add a new local check script
+- Improve pattern list in check_claims.py
+- Add a metric naming check
diff --git a/docs/evidence/GUIDE_ALLOWED_CLAIMS.md b/docs/evidence/GUIDE_ALLOWED_CLAIMS.md
new file mode 100644
index 00000000..eb4809f3
--- /dev/null
+++ b/docs/evidence/GUIDE_ALLOWED_CLAIMS.md
@@ -0,0 +1,18 @@
+# Allowed Public Claim Examples
+
+Examples of allowed and forbidden public claims.
+
+## Allowed
+- "Computationally nominated 24 antimicrobial peptide candidates."
+- "Pipeline achieves AUROC 0.78 on benchmark — limitations apply."
+- "All candidates are dry-lab predictions requiring wet-lab validation."
+- "Full reproducible evidence trail available for scientific review."
+
+## Forbidden
+- "Discovered a new antibiotic." (Requires wet-lab validation)
+- "Pipeline-validated antimicrobial." (Computation does not validate)
+- "Safe, effective peptide therapy." (No clinical evidence)
+- "AI drug discovery breakthrough." (Overclaiming)
+
+## Rule
+When in doubt, use the weakest accurate claim. Add a caveat.
diff --git a/docs/evidence/GUIDE_DECISION_VALIDATOR.md b/docs/evidence/GUIDE_DECISION_VALIDATOR.md
new file mode 100644
index 00000000..27476bdf
--- /dev/null
+++ b/docs/evidence/GUIDE_DECISION_VALIDATOR.md
@@ -0,0 +1,18 @@
+# Decision Record Template Validator
+
+Validates decision log entries against the schema.
+
+## Usage
+```bash
+python -m openamp_foundry.cli validate \
+  --certificate decision_log_entry.json \
+  --schema schemas/decision_log.schema.json
+```
+
+## Checks
+The schema validates:
+- Required fields are present
+- Decision type is one of the allowed values
+- Decision is approved, rejected, or deferred
+- If dissent_flag is true, dissent_notes are required
+- Date is in ISO 8601 format
diff --git a/docs/evidence/GUIDE_ISSUE_TRIAGE.md b/docs/evidence/GUIDE_ISSUE_TRIAGE.md
new file mode 100644
index 00000000..653ab4f1
--- /dev/null
+++ b/docs/evidence/GUIDE_ISSUE_TRIAGE.md
@@ -0,0 +1,16 @@
+# Issue Triage Matrix
+
+How to triage incoming issues.
+
+| Factor | High Priority | Medium Priority | Low Priority |
+|--------|:-------------:|:---------------:|:------------:|
+| Safety impact | Direct safety concern | Indirect safety concern | No safety impact |
+| User impact | Blocks all users | Blocks some users | Minor inconvenience |
+| Effort | Small fix (< 1 day) | Medium fix (1-3 days) | Large fix (> 3 days) |
+| Alignment | Core mission | Supporting | Nice-to-have |
+
+## Process
+1. Apply appropriate labels (bug, enhancement, docs, safety).
+2. Assign priority (P0-P3).
+3. If P0, notify maintainers immediately.
+4. If good-first-issue, add the label.
diff --git a/docs/evidence/GUIDE_PLATFORM_VERIFICATION.md b/docs/evidence/GUIDE_PLATFORM_VERIFICATION.md
new file mode 100644
index 00000000..0eb26f1f
--- /dev/null
+++ b/docs/evidence/GUIDE_PLATFORM_VERIFICATION.md
@@ -0,0 +1,15 @@
+# Platform Quickstart Verification Notes
+
+Quickstart verification results by platform.
+
+| Platform | Status | Notes |
+|----------|:------:|-------|
+| macOS (Intel) | ✅ | Fully supported |
+| macOS (Apple Silicon) | ✅ | Fully supported (native) |
+| Ubuntu 22.04 | ✅ | Fully supported |
+| Ubuntu 24.04 | ✅ | Fully supported |
+| Windows (WSL2) | ⚠️ | Works with WSL2, not tested natively |
+| Windows (native) | ❌ | Not tested — use WSL2 |
+
+## If Something Doesn't Work
+Check the troubleshooting guide: `docs/evidence/GUIDE_QUICKSTART_DIAGNOSIS.md`
diff --git a/docs/evidence/GUIDE_REPRODUCIBILITY_CHECK.md b/docs/evidence/GUIDE_REPRODUCIBILITY_CHECK.md
new file mode 100644
index 00000000..ebac84fb
--- /dev/null
+++ b/docs/evidence/GUIDE_REPRODUCIBILITY_CHECK.md
@@ -0,0 +1,22 @@
+# Hidden-State Reproducibility Check
+
+How to verify that pipeline results are reproducible.
+
+## What to Check
+1. Same inputs + same config + same version = same outputs.
+2. Output hashes match between runs.
+3. Random seeds produce deterministic results.
+
+## How to Verify
+```bash
+# Run twice and compare outputs
+make demo
+cp outputs/demo_ranked.jsonl /tmp/run1.jsonl
+make demo
+diff outputs/demo_ranked.jsonl /tmp/run1.jsonl  # Should be identical
+```
+
+## What If Outputs Differ
+- Check for non-deterministic components (timestamps, UUIDs).
+- Check that random seeds are fixed.
+- Check that the code version hasn't changed between runs.
diff --git a/docs/evidence/GUIDE_REVIEW_DASHBOARD.md b/docs/evidence/GUIDE_REVIEW_DASHBOARD.md
new file mode 100644
index 00000000..3b77566f
--- /dev/null
+++ b/docs/evidence/GUIDE_REVIEW_DASHBOARD.md
@@ -0,0 +1,15 @@
+# Maintainer Review Dashboard
+
+Metrics for tracking review performance.
+
+## Dashboard Metrics
+| Metric | Target | Collection |
+|--------|:------:|------------|
+| Open PRs | < 10 | `gh pr list` |
+| PRs awaiting review | < 5 | `gh pr list --review-requested` |
+| Median time to review | < 24h | Tracked in review latency ledger |
+| Oldest unreviewed PR | < 72h | `gh pr list --sort created` |
+| Issues without assignee | < 20 | `gh issue list --no-assignee` |
+
+## How to Use
+Run weekly. If any metric exceeds target, investigate bottlenecks.
diff --git a/docs/evidence/INDEX_ADR.md b/docs/evidence/INDEX_ADR.md
new file mode 100644
index 00000000..0cc9d0e6
--- /dev/null
+++ b/docs/evidence/INDEX_ADR.md
@@ -0,0 +1,19 @@
+# Architecture Decision Record Index
+
+Index of architecture decision records.
+
+| ADR | Title | Date | Status |
+|:---:|-------|:----:|:------:|
+| 001 | Use heuristic scoring over ML | 2026-06 | Accepted |
+| 002 | Dry-lab first, wet-lab gated | 2026-06 | Accepted |
+| 003 | JSON schemas for all artifacts | 2026-06 | Accepted |
+| 004 | 50-loop execution plan | 2026-06 | Accepted |
+| 005 | Simulation modules experimental | 2026-07 | Accepted |
+
+## Format
+ADRs are stored in `docs/evidence/` with the prefix `ADR-`.
+
+## Rules
+- Major architecture decisions should be recorded as ADRs.
+- ADRs are numbered sequentially.
+- ADRs are never deleted — only superseded.
diff --git a/docs/evidence/MATRIX_CONTRIBUTOR_ENV.md b/docs/evidence/MATRIX_CONTRIBUTOR_ENV.md
new file mode 100644
index 00000000..86d06408
--- /dev/null
+++ b/docs/evidence/MATRIX_CONTRIBUTOR_ENV.md
@@ -0,0 +1,21 @@
+# Contributor Environment Matrix
+
+Supported environments for contributors.
+
+| OS | Python | Shell | Status |
+|---|:-------:|:-----:|:------:|
+| macOS 14+ | 3.11-3.14 | zsh | ✅ |
+| macOS 13 | 3.11-3.14 | zsh | ✅ |
+| Ubuntu 22.04+ | 3.11-3.14 | bash | ✅ |
+| Ubuntu 20.04 | 3.11-3.13 | bash | ✅ |
+| Windows (WSL2/Ubuntu) | 3.11-3.13 | bash | ⚠️ |
+
+## Requirements
+- Python 3.11+ (3.14 recommended)
+- pip 21+
+- git 2.25+
+- make (optional, for Makefile targets)
+
+## Notes
+- Windows native is not supported — use WSL2.
+- Other Linux distributions likely work but are not tested in CI.
diff --git a/tests/test_batch_v2.py b/tests/test_batch_v2.py
new file mode 100644
index 00000000..f357ab52
--- /dev/null
+++ b/tests/test_batch_v2.py
@@ -0,0 +1,20 @@
+"""Tests for latest batch."""
+from pathlib import Path
+
+FILES = [
+    "CATALOG_AGENT_ISSUES.md",
+    "GUIDE_ALLOWED_CLAIMS.md",
+    "GUIDE_ISSUE_TRIAGE.md",
+    "GUIDE_REVIEW_DASHBOARD.md",
+    "INDEX_ADR.md",
+    "GUIDE_DECISION_VALIDATOR.md",
+    "GUIDE_PLATFORM_VERIFICATION.md",
+    "MATRIX_CONTRIBUTOR_ENV.md",
+    "GUIDE_REPRODUCIBILITY_CHECK.md",
+    "ARCHITECTURE_OVERVIEW_CONTRIBUTORS.md",
+]
+
+
+def test_all_exist():
+    for f in FILES:
+        assert Path("docs/evidence", f).exists(), f"Missing: {f}"