Add 10 decision trees, walkthroughs, and runbooks

docs/evidence/DECISION_TREE_BENCHMARK_CHANGE.md

@@ -0,0 +1,11 @@
+# Decision Tree: Benchmark Change Review
+
+```mermaid
+flowchart TD
+  A[Benchmark change proposed] --> B{Does it change AUROC?}
+  B -->|Yes, improvement| C[Verify no overfitting]
+  B -->|Yes, regression| D[Is regression > 0.02?]
+  D -->|Yes| E[Block — investigate]
+  D -->|No| F[Accept — document tradeoff]
+  B -->|No change| G[Verify benchmark still runs]
+```

docs/evidence/DECISION_TREE_EVIDENCE_DOWNGRADE.md

@@ -0,0 +1,12 @@
+# Decision Tree: Evidence Downgrade
+
+```mermaid
+flowchart TD
+  A[Evidence quality concern] --> B{Is the evidence reproducible?}
+  B -->|No| C[Downgrade 1 level]
+  B -->|Yes| D{Are the caveats documented?}
+  D -->|No| E[Downgrade 1 level]
+  D -->|Yes| F{Is the data source reliable?}
+  F -->|No| G[Downgrade 1 level]
+  F -->|Yes| H[Maintain current level]
+```

docs/evidence/DECISION_TREE_FIRST_CONTRIBUTION.md

@@ -0,0 +1,12 @@
+# Decision Tree: Choosing First Contribution
+
+```mermaid
+flowchart TD
+  A[Want to contribute?] --> B{Can you run Python?}
+  B -->|Yes| C{Can you write tests?}
+  B -->|No| D[Start with docs-only tasks]
+  C -->|Yes| E[Pick a test issue]
+  C -->|No| F{Can you write docs?}
+  F -->|Yes| G[Pick a docs issue]
+  F -->|No| H[Start with CONTRIBUTOR_CURRICULUM.md]
+```

docs/evidence/DECISION_TREE_OPENING_DOCS_PR.md

@@ -0,0 +1,12 @@
+# Decision Tree: Opening a Documentation PR
+
+```mermaid
+flowchart TD
+  A[Change needed in docs] --> B{Is it a typo or formatting?}
+  B -->|Yes| C[Direct commit or minor PR]
+  B -->|No| D{Is it a policy change?}
+  D -->|Yes| E[Full PR + maintainer review]
+  D -->|No| F{Is it a new document?}
+  F -->|Yes| G[PR + link from PROJECT_INDEX.md]
+  F -->|No| H[Standard PR with review]
+```

docs/evidence/DECISION_TREE_PUBLIC_SHARING.md

@@ -0,0 +1,12 @@
+# Decision Tree: Public Artifact Sharing
+
+```mermaid
+flowchart TD
+  A[Share artifact publicly?] --> B{Contains sensitive data?}
+  B -->|Yes| C[Redact or don't share]
+  B -->|No| D{Is it a release artifact?}
+  D -->|Yes| E[Follow PUBLICATION_PACK.md]
+  D -->|No| F{Is it a benchmark result?}
+  F -->|Yes| G[Include caveats]
+  F -->|No| H[Share with standard disclaimer]
+```

docs/evidence/GUIDE_RELEASE_READINESS.md

@@ -0,0 +1,16 @@
+# Decision Guide: Release Readiness
+
+Before releasing, verify:
+
+- [ ] `make full-reproducibility-report` succeeds
+- [ ] `python scripts/check_doc_links.py` — 0 broken links
+- [ ] `python scripts/safety/check_claims.py` — 0 overclaiming
+- [ ] `python3 -m pytest -q` — all tests pass
+- [ ] METRICS_CURRENT.md is up to date
+- [ ] PUBLICATION_PACK.md checklist is complete
+- [ ] Decision log is current
+
+## Decision
+- All checks pass → Release candidate
+- Minor issues → Fix before release
+- Major issues → Block release

docs/evidence/RUNBOOK_BROKEN_DOCS_LINKS.md

@@ -0,0 +1,9 @@
+# Maintainer Runbook: Broken Docs Links
+
+When doc links are broken:
+
+1. Run `python scripts/check_doc_links.py` to find all broken links.
+2. For each broken link, determine the correct target path.
+3. Update the link in the source document.
+4. If the target document was moved, add a redirect or update all references.
+5. Re-run the check to confirm 0 broken links.

docs/evidence/RUNBOOK_STALE_ARTIFACTS.md

@@ -0,0 +1,10 @@
+# Maintainer Runbook: Stale Artifacts
+
+When an artifact is stale:
+
+1. Identify what's stale and why.
+2. If it's a benchmark output, regenerate with `make bench-*`.
+3. If it's a schema, update it to match current code.
+4. If it's a config, update or document the discrepancy.
+5. Update references in docs.
+6. Add a test to prevent future staleness.

docs/evidence/RUNBOOK_UNSAFE_CLAIMS.md

@@ -0,0 +1,10 @@
+# Maintainer Runbook: Unsafe Claim Wording
+
+When unsafe claim wording is found:
+
+1. Run `python scripts/safety/check_claims.py` to see all flagged phrases.
+2. Review each flagged phrase in context.
+3. Rewrite overclaiming phrases using project-allowed terms.
+4. If the claim is in a public-facing document, prioritize the fix.
+5. Re-run the check to confirm the fix.
+6. If the author disagrees, escalate to maintainers.

docs/evidence/WALKTHROUGH_CLAIM_TO_EVIDENCE.md

@@ -0,0 +1,16 @@
+# Guided Walkthrough: Claim-to-Evidence Matrices
+
+How to map a claim to its supporting evidence.
+
+## Steps
+1. State the claim clearly — what exactly is being asserted?
+2. Identify the proof level required (see PROOF_LADDER.md).
+3. Find the evidence that supports the claim.
+4. Check if the evidence level matches the claim strength.
+5. If the claim is stronger than the evidence, weaken the claim or gather more evidence.
+
+## Example
+| Claim | Required Level | Available Evidence | Match? |
+|-------|:--------------:|-------------------|:------:|
+| "Candidate X is predicted antimicrobial" | 2 (baseline score) | Ensemble score 0.85 | ✅ |
+| "Candidate X is a drug candidate" | 5 (lab assay) | No wet-lab data | ❌ |

tests/test_decision_trees.py

@@ -0,0 +1,20 @@
+"""Tests for batch of 10 docs."""
+from pathlib import Path
+
+FILES = [
+    "WALKTHROUGH_CLAIM_TO_EVIDENCE.md",
+    "DECISION_TREE_FIRST_CONTRIBUTION.md",
+    "DECISION_TREE_OPENING_DOCS_PR.md",
+    "DECISION_TREE_BENCHMARK_CHANGE.md",
+    "GUIDE_RELEASE_READINESS.md",
+    "DECISION_TREE_PUBLIC_SHARING.md",
+    "DECISION_TREE_EVIDENCE_DOWNGRADE.md",
+    "RUNBOOK_STALE_ARTIFACTS.md",
+    "RUNBOOK_BROKEN_DOCS_LINKS.md",
+    "RUNBOOK_UNSAFE_CLAIMS.md",
+]
+
+
+def test_all_exist():
+    for f in FILES:
+        assert Path("docs/evidence", f).exists(), f"Missing: {f}"