fix(skills): preserve superseded PR attribution

.agents/skills/nemoclaw-maintainer-find-review-pr/SKILL.md

@@ -63,11 +63,39 @@ gh pr view <number> --json number,title,author,createdAt,additions,deletions,rev
 
 ## Step 4: Check for superseded PRs
 
-Flag a PR when its body contains one of these phrases:
+Run the comparator's canonical detector with the open candidate PR numbers:
 
-- `follow-up to #NNN` / `supersedes #NNN` / `replaces #NNN` / `folds in #NNN`
+```bash
+../nemoclaw-maintainer-pr-comparator/scripts/parse-supersession.sh <pr-number-1> <pr-number-2> ...
+```
+
+It recognizes the comparator parser's case-insensitive statement families:
+
+- `supersed[a-z]*` before `#N`: `supersedes #N` points from the current PR to `#N`; `superseded by #N` points from `#N` to the current PR.
+- `replac[a-z]*` before `#N`: `replaces #N` points from the current PR to `#N`; `replaced by #N` points from `#N` to the current PR.
+- `clos[a-z]* in favor of` before `#N`: `closes in favor of #N` and `closed in favor of #N` point from `#N` to the current PR.
+- `fold[a-z]* in` before `#N`: `folds in #N` points from the current PR to `#N`; `folded into #N` points from `#N` to the current PR.
+
+The bracket expressions describe the parser grammar; they are not literal PR body text.
+A `follow-up to #N` statement is a related-PR signal, not a supersession declaration,
+unless one of these phrases also appears.
+
+Each supersession phrase must name another open candidate PR. It indicates that one PR can include the other.
+It records a relationship but does not prove that the target carries the source PR's work.
+
+When the target claims the source PR's full scope, compare their commits and diffs.
+If material code, tests, or documentation from another contributor remains in the target,
+apply the canonical policy in
+`../nemoclaw-maintainer-policies/references/workflow-policy.md`.
+
+This skill reports recommendations only.
+Do not recommend closing the source PR until another authorized workflow has:
 
-The phrase must name another open candidate PR. It indicates that one PR can include the other.
+- completed any required transfer
+- verified the updated commits, attribution, and CI
+- rerun the comparator and selected the target
+- confirmed that the target contains the source's full scope and required contributor attribution
+- merged the selected target
 
 ## Step 5: Present results
 
@@ -92,7 +120,8 @@ For superseded PRs:
 ### Superseded PRs
 
 - #1416 supersedes/folds in #1392 (shell-quote sandboxName)
-  Consider closing #1392 if #1416 contains its full scope.
+  Keep #1392 open while an authorized workflow completes any required transfer, verifies the updated commits, attribution, and CI, and reruns the comparator.
+  After the updated verdict selects #1416 and #1416 merges, consider closing #1392 only if #1416 contains its full scope and preserves any required contributor attribution.
 ```
 
 ### Clean candidates

.agents/skills/nemoclaw-maintainer-policies/references/workflow-policy.md

@@ -46,6 +46,40 @@ The contributor must correct a failure.
 Maintainers must reject a PR with an unverified commit or no DCO declaration.
 Do not merge, approve, or repair it for the contributor.
 
+## Superseded PR Attribution
+
+A supersession declaration records a relationship between PRs.
+It does not prove that one PR covers the other, satisfies contributor requirements, or ranks above another candidate.
+
+When a replacement PR carries material code, tests, or documentation from another contributor's PR:
+
+- Confirm that the source PR already contains that contributor's `Signed-off-by:` declaration.
+  Never add or copy a DCO declaration on another contributor's behalf.
+- Use the exact author name and email from the source commit.
+  Never guess or substitute an attribution identity.
+
+If the source commit has no usable author identity, leave the winner unset and ask the contributor to provide machine-readable attribution.
+If the source PR has no contributor DCO declaration, leave the winner unset and ask the contributor to add it.
+A maintainer must not supply either declaration on the contributor's behalf.
+
+After both checks pass:
+
+- Add `Supersedes #<number>` to the replacement PR body and identify the transferred contribution.
+- Preserve the source contributor as the Git author when cherry-picking their commit.
+- When the replacement combines or reconstructs their work, add a
+  `Co-authored-by: Name <email>` commit trailer for each contributor whose work remains.
+  Use the verified source-commit identity.
+- Keep the replacement author's own DCO declaration in the replacement PR body.
+- Before recommending merge or closure, verify the replacement PR body and commit metadata,
+  and confirm that every replacement commit appears as `Verified` in GitHub.
+
+An independent implementation based only on the issue, a reproduction, or public discussion does not require co-authorship.
+Still link the related PRs when recommending closure so the decision remains discoverable.
+A comment on the superseded PR does not replace attribution in the merged PR history.
+
+This policy does not authorize a merge, close, comment, or other write.
+Each operation still requires its own authorization.
+
 ## Issue Classification
 
 Native GitHub Issue Type is the canonical issue-kind field:

.agents/skills/nemoclaw-maintainer-pr-comparator/SKILL.md

@@ -20,6 +20,8 @@ Tier 3 resolves ties. If no PR passes Tier 0, rank eligible PRs for salvage.
 ## Repo policy
 
 The defaults use NemoClaw conventions for CODEOWNERS, DCO, CodeRabbit, and `docs/`.
+Read the canonical superseded-PR attribution policy in
+`../nemoclaw-maintainer-policies/references/workflow-policy.md`.
 Edit `repo-policy.md` for another repository.
 
 ## Workflow
@@ -30,7 +32,7 @@ Copy this checklist into your response and check off each step:
 PR Comparison Progress:
 - [ ] Step 1: Parse issue (body + comments) for acceptance criteria
 - [ ] Step 2: Discover candidate PRs in the defined order
-- [ ] Step 3: Detect supersession (parse PR bodies)
+- [ ] Step 3: Detect supersession and classify transferred work
 - [ ] Step 4: Run Tier 0 gates per PR
 - [ ] Step 5: Run Tier 1 correctness checks per PR
 - [ ] Step 6: Run Tier 2 quality checks per PR
@@ -57,14 +59,32 @@ scripts/find-candidates.sh <issue-number>
 
 Applies a single default order with stop conditions.
 
-### Step 3: Detect supersession
+### Step 3: Detect supersession and transferred work
 
 ```bash
 scripts/parse-supersession.sh <pr-number-1> <pr-number-2> ...
 ```
 
-Parse these statements from each PR body: `supersedes #N`, `replaces #N`, `closes in favor of #N`, and `folds in #N`.
-Use supersession as the first tiebreaker.
+Parse the case-insensitive statement families implemented by `scripts/parse-supersession.sh`:
+
+- `supersed[a-z]*` before `#N`: `supersedes #N` points from the current PR to `#N`; `superseded by #N` points from `#N` to the current PR.
+- `replac[a-z]*` before `#N`: `replaces #N` points from the current PR to `#N`; `replaced by #N` points from `#N` to the current PR.
+- `clos[a-z]* in favor of` before `#N`: `closes in favor of #N` and `closed in favor of #N` point from `#N` to the current PR.
+- `fold[a-z]* in` before `#N`: `folds in #N` points from the current PR to `#N`; `folded into #N` points from `#N` to the current PR.
+
+The bracket expressions describe the parser grammar; they are not literal PR body text.
+A `follow-up to #N` statement is a related-PR signal, not a supersession declaration, unless one of these phrases also appears.
+These statements record a relationship.
+They do not rank a candidate or prove that its diff contains another contributor's work.
+
+For each declared or suspected replacement, compare the commits and diffs and classify the relationship:
+
+- `independent`: The PR implements the issue without carrying material code, tests, or documentation from another contributor.
+- `transferred`: The PR carries material work from another contributor.
+- `unclear`: The available evidence does not establish whether another contributor's work remains.
+
+For `transferred`, apply the canonical superseded-PR attribution policy before setting a winner or recommending that the source PR be closed.
+For `unclear`, leave `winner` null and request maintainer judgment.
 
 ### Step 4: Tier 0 gates
 
@@ -102,6 +122,7 @@ Apply the four model checks in `checks/tier-2-quality.md`.
 Compute the mode from the Tier 0 results. Do not accept a mode from the caller.
 In happy mode, set `winner` only to an eligible PR and set `closest_to_ready` to null.
 Leave `winner` null when the evidence does not support a merge recommendation.
+Do not set `winner` for a replacement with transferred work until the required attribution is present and verified.
 In degraded mode, set `winner` to null.
 Set `closest_to_ready` only to an open PR that passes contributor requirements.
 See `tiebreakers.md`.

.agents/skills/nemoclaw-maintainer-pr-comparator/scripts/parse-supersession.sh

@@ -5,8 +5,10 @@
 # Parse PR bodies for supersession references and emit edges:
 #   <superseder_pr> -> <superseded_pr>
 #
-# Patterns matched (case-insensitive): "supersedes #N", "replaces #N",
-# "closes in favor of #N", "closed in favor of #N", "folds in #N".
+# Patterns matched case-insensitively. Active forms such as "supersedes #N",
+# "replaces #N", and "folds in #N" point from the current PR to #N. Passive
+# forms such as "superseded by #N", "replaced by #N", "closed in favor of
+# #N", and "folded into #N" point from #N to the current PR.
 #
 # Usage: parse-supersession.sh <pr-1> <pr-2> [...] [--repo OWNER/REPO]
 
@@ -36,19 +38,34 @@ done
 candidates_set=$(printf '%s\n' "${prs[@]}" | sort -u)
 
 edges=()
+append_edge() {
+  local candidate="$1"
+  local existing
+
+  for existing in "${edges[@]}"; do
+    [ "$existing" = "$candidate" ] && return
+  done
+  edges+=("$candidate")
+}
+
+reverse_pattern='(supersed[a-z]*[[:space:]]+by|replac[a-z]*[[:space:]]+by|clos[a-z]*[[:space:]]+in[[:space:]]+favor[[:space:]]+of|fold[a-z]*[[:space:]]+into)'
 for pr in "${prs[@]}"; do
   body=$(gh pr view "$pr" "${repo_args[@]}" --json body --jq .body 2>/dev/null || echo "")
   [ -z "$body" ] && continue
 
-  # Extract referenced PR numbers from supersession patterns.
-  while IFS= read -r ref; do
+  # Extract supersession statements and orient passive forms toward this PR.
+  while IFS= read -r statement; do
+    ref="${statement##*#}"
     # Only emit edges where the target is also a candidate.
     if printf '%s\n' "$candidates_set" | grep -q "^${ref}$"; then
-      edges+=("$pr -> $ref")
+      normalized=$(printf '%s' "$statement" | tr '[:upper:]' '[:lower:]')
+      if [[ $normalized =~ $reverse_pattern ]]; then
+        append_edge "$ref -> $pr"
+      else
+        append_edge "$pr -> $ref"
+      fi
     fi
   done < <(printf '%s' "$body" | grep -oiE '(supersed[a-z]*|replac[a-z]*|clos[a-z]* in favor of|fold[a-z]* in)[^#]*#([0-9]+)' \
-    | grep -oE '#[0-9]+' \
-    | tr -d '#' \
     | sort -u)
 done
 

.agents/skills/nemoclaw-maintainer-pr-comparator/templates/verdict.md

@@ -46,23 +46,31 @@ Render the final scorecard with `scripts/render-verdict.py`. Below is the human-
 | <criterion 2> | covered | missing |
 | <criterion 3> | missing | covered |
 
-### Verdict: MERGE PR #A
+### Verdict: No clear winner — see scorecard for recommended action
 
 Reasoning trace:
 - PR #B failed Tier 0. Check `<name>` failed on `<short-sha>` after the force-push.
 - PR #A scored 14.5. PR #B scored 9.0.
-- PR #A misses criterion 3. Cherry-pick PR #B's test at `<file>:<line>` to cover it.
+- PR #A misses criterion 3. PR #B contains the test at `<file>:<line>`, so PR #A needs a transfer before it can be selected.
 
 ### Suggested action
 
-1. Merge PR #A.
-2. Cherry-pick the test from PR #B at `<file>:<line-range>` to cover criterion 3.
-3. Close PR #B with a comment that links to #A and records the cherry-pick.
+1. Confirm that PR #B contains the contributor's `Signed-off-by:` declaration. Do not add or copy that declaration on the contributor's behalf. If it is absent, leave the winner unset and ask the contributor.
+2. Read the exact author name and email from the source commit. Never guess or substitute it. If no usable identity exists, leave the winner unset and ask the contributor.
+3. Transfer the test from PR #B before merge. Prefer `git cherry-pick -S -x <source-sha>` so the source contributor remains the Git author.
+4. If the work must be combined or reconstructed, add a `Co-authored-by: Name <email>` trailer using the verified source-commit identity.
+5. Add `Supersedes #B` to PR #A's body and identify the transferred test.
+6. Keep the replacement author's own DCO declaration in PR #A's body.
+7. Confirm that every replacement commit appears as `Verified` in GitHub.
+8. Verify the updated commits, attribution, and CI, then run the comparator again on the updated SHA.
+9. Merge PR #A only if the new verdict selects it.
+10. After PR #A merges, close PR #B with a comment that links to #A.
 
 ### Reasoning evidence
 - CI: all 12 required checks passed on PR #A commit `<short-sha>`. On PR #B commit `<short-sha>`, `test-cli` failed at `<log-line>`.
 - Tier 1.1 PR #A: The test at `<file>:<line-range>` asserts on `<output>`. The previous code returned `<wrong-output>`, so the assertion would have failed.
 - Tier 1.3 PR #A fail: no test for empty-input edge case despite issue commenter raising it at `issue.comment.4`
+- Attribution: PR #A carries `<contribution>` from PR #B. Its body names the source PR, and commit `<sha>` preserves `<contributor>` as Git author or co-author.
 - ... <one entry per judgment> ...
 ```
 

.agents/skills/nemoclaw-maintainer-pr-comparator/tiebreakers.md

@@ -23,13 +23,18 @@ Eliminate any PR failing Tier 0. Among survivors:
    Set the winner when the evidence distinguishes a PR.
    Otherwise, leave `winner` null.
 
+### Supersession relationship
+
+A supersession statement records a relationship between candidates.
+It does not prove coverage or attribution, and it does not rank a candidate.
+Classify transferred work and complete the canonical attribution checks before recommending the replacement PR.
+
 ### Tiebreakers (in order)
 
-1. **Supersession.** Prefer a PR whose body states that it supersedes another candidate. See `scripts/parse-supersession.sh`.
-2. **Smaller diff.** Prefer the smaller diff when both PRs cover the issue scope.
-3. **Better edge-case test coverage.** Compare Tier 1.3 (negative test coverage) outputs.
-4. **Most recent activity.** Prefer the PR with the most recent commit.
-5. **Lower PR number.** Use the lower PR number if the PRs remain tied.
+1. **Smaller diff.** Prefer the smaller diff when both PRs cover the issue scope.
+2. **Better edge-case test coverage.** Compare Tier 1.3 (negative test coverage) outputs.
+3. **Most recent activity.** Prefer the PR with the most recent commit.
+4. **Lower PR number.** Use the lower PR number if the PRs remain tied.
 
 ## Degraded mode (no PR passes Tier 0)
 
@@ -68,5 +73,6 @@ For each acceptance criterion (from issue body + comments), build a row showing
 
 Use the matrix to find tests or changes that the selected PR does not include.
 The verdict can recommend a small transfer from another PR.
+Complete the transfer and required attribution, then rerun the comparator before selecting a winner.
 
 Per-criterion winner cells: `covered` (full), `partial` (yellow), `missing` (red).