docs(vulnogram): oss-security is sent automatically by the OSS/ASF Emails tab

skills/pr-management-triage/SKILL.md

@@ -442,6 +442,13 @@ notified:
   blocks every other. See
   [`tools/agent-guard`](../../tools/agent-guard/README.md) and
   [`comment-templates.md`](comment-templates.md#the-folded-maintainer-triage-note--the-single-contributor-channel).
+- Exemption — **your own PR/issue**: this rule targets triaging
+  *other* people's PRs. When the operator is themselves the author
+  (author == the authenticated `gh` user), the guard allows
+  `@`-mentioning maintainers/reviewers — nudging your own reviewers
+  from your own PR is a legitimate, deliberate act. A one-off
+  `MAGPIE_ALLOW_MENTIONS=1` override is the escape hatch for any
+  other intentional exception.
 
 This supersedes Golden rule 9's "pings still notify a maintainer"
 expectation for the operator/reviewer side: F5a/F5b still make the

skills/pr-management-triage/guards/mention.py

@@ -23,7 +23,13 @@
 author-directed comments (`gh pr/issue comment`) — in both, the author's
 @-mention is permitted (it is the intended "your move" signal) and any other
 @-mention (a maintainer: operator, reviewer, CODEOWNER, team) is blocked.
-Maintainer handles must be backtick-quoted so they never notify. Discovered by
+Two exceptions lift the block: an explicit ``MAGPIE_ALLOW_MENTIONS=1`` override
+(a deliberate one-off, requested by the operator), and the operator commenting
+on their **own** PR/issue — when the target's author is the authenticated ``gh``
+user, mentioning maintainers is a legitimate self-directed nudge to one's own
+reviewers, not the drive-by maintainer spam this guard exists to stop.
+Otherwise maintainer handles must be backtick-quoted so they never notify.
+Discovered by
 the agent-guard PreToolUse dispatcher from a guards.d directory — see
 tools/agent-guard for the engine and the GuardContext API. Import-free:
 everything comes from ``ctx``.
@@ -70,6 +76,13 @@ def guard(ctx):
             "author is known, drop the @-mentions (use backticked `login`), or override "
             "with MAGPIE_ALLOW_MENTIONS=1 if the mention is intentional."
         )
+    # Operator's own PR/issue: when the target's author is the authenticated gh
+    # user, mentioning maintainers is a self-directed nudge to one's own reviewers
+    # (legitimate), not the drive-by maintainer spam this guard blocks. Resolution
+    # failing falls through to the normal author-only rule (safe default).
+    operator = ctx.run(["gh", "api", "user", "--jq", ".login"])
+    if operator and operator.lower() == author.lower():
+        return None
     offenders = sorted({m for m in mentions if m != author.lower()})
     if offenders:
         return (

tools/agent-guard/README.md

@@ -64,7 +64,7 @@ way — see [Contributing guards](#contributing-guards)):
 
 | Guard | Owner skill | Blocks | Rule it enforces |
 |---|---|---|---|
-| `mention` | `pr-management-triage` | `gh pr comment` / `gh issue comment` that `@`-mentions anyone other than the PR/issue author; **any** `@`-mention in `gh pr edit --body[-file]` | denoise: author-directed feedback never pings maintainers; body edits stay silent |
+| `mention` | `pr-management-triage` | `gh pr comment` / `gh issue comment` that `@`-mentions anyone other than the PR/issue author; **any** `@`-mention in `gh pr edit --body[-file]` | denoise: author-directed feedback never pings maintainers; body edits stay silent. Exempt: the operator commenting on their **own** PR/issue (author == authenticated `gh` user), and the `MAGPIE_ALLOW_MENTIONS=1` override |
 | `mark-ready` | `pr-management-triage` | adding `ready for maintainer review` while the PR head SHA has GitHub Actions runs awaiting approval | Golden rule 1b |
 | `security-language` | `security-issue-fix` | a CVE id / security-fix language in a **public** `gh pr create`/`gh pr edit` title/body (not comments) | public-PR scrubbing |
 

tools/agent-guard/tests/test_skill_guards.py

@@ -51,6 +51,19 @@ def _stub(args, cwd=None):
     return _stub
 
 
+def gh_stub(*, author, operator="operator-bot"):
+    """Distinguish the guard's two lookups: ``gh api user`` (operator identity)
+    vs. ``gh pr/issue view`` (target author). Default operator differs from the
+    author so a bare stub keeps the author-only rule in force."""
+
+    def handler(args):
+        if "api" in args and "user" in args:
+            return operator
+        return author
+
+    return handler
+
+
 def dispatch(command):
     return agent_guard.dispatch(command, cwd=None)
 
@@ -65,17 +78,39 @@ def test_guard_files_exist():
 
 
 def test_mention_author_allowed(monkeypatch):
-    monkeypatch.setattr(agent_guard, "_run", fake_run(lambda a: "alice"))
+    monkeypatch.setattr(agent_guard, "_run", fake_run(gh_stub(author="alice")))
     assert dispatch('gh pr comment 5 --body "@alice thanks"') is None
 
 
 def test_mention_non_author_denied(monkeypatch):
-    monkeypatch.setattr(agent_guard, "_run", fake_run(lambda a: "alice"))
+    monkeypatch.setattr(agent_guard, "_run", fake_run(gh_stub(author="alice")))
     reason = dispatch('gh pr comment 5 --body "@bob please review"')
     assert reason and "bob" in reason and "mention" in reason
 
 
-def test_fold_any_mention_denied():
+def test_mention_own_pr_allows_maintainers(monkeypatch):
+    # Operator commenting on their own PR may @-mention maintainers/reviewers.
+    monkeypatch.setattr(agent_guard, "_run", fake_run(gh_stub(author="alice", operator="alice")))
+    assert dispatch('gh pr comment 5 --body "@bob @carol please take a look"') is None
+
+
+def test_mention_own_pr_case_insensitive(monkeypatch):
+    monkeypatch.setattr(agent_guard, "_run", fake_run(gh_stub(author="Alice", operator="alice")))
+    assert dispatch('gh pr comment 5 --body "@bob review please"') is None
+
+
+def test_mention_others_pr_still_denied_when_operator_known(monkeypatch):
+    # Author is someone else; operator resolving successfully must not relax the rule.
+    monkeypatch.setattr(agent_guard, "_run", fake_run(gh_stub(author="alice", operator="carol")))
+    reason = dispatch('gh pr comment 5 --body "@bob ping"')
+    assert reason and "bob" in reason and "mention" in reason
+
+
+def test_fold_any_mention_denied(monkeypatch):
+    # Hermetic: a non-author maintainer @-mention in a fold edit is denied.
+    # Stubbed so it never depends on a real `gh pr view` (and so the own-PR
+    # exemption, author == operator, is not accidentally triggered).
+    monkeypatch.setattr(agent_guard, "_run", fake_run(gh_stub(author="bob", operator="carol")))
     reason = dispatch('gh pr edit 5 --body "@alice heads up"')
     assert reason and "fold" in reason
 
@@ -85,7 +120,7 @@ def test_fold_clean_allowed():
 
 
 def test_mention_body_file(monkeypatch, tmp_path):
-    monkeypatch.setattr(agent_guard, "_run", fake_run(lambda a: "alice"))
+    monkeypatch.setattr(agent_guard, "_run", fake_run(gh_stub(author="alice")))
     body = tmp_path / "b.md"
     body.write_text("@bob please look", encoding="utf-8")
     assert dispatch(f"gh pr comment 5 --body-file {body}") is not None
@@ -98,7 +133,7 @@ def test_mention_author_unresolved_fails_closed(monkeypatch):
 
 
 def test_mention_override(monkeypatch):
-    monkeypatch.setattr(agent_guard, "_run", fake_run(lambda a: "alice"))
+    monkeypatch.setattr(agent_guard, "_run", fake_run(gh_stub(author="alice")))
     assert dispatch('MAGPIE_ALLOW_MENTIONS=1 gh pr comment 5 --body "@bob ping"') is None
 
 

tools/cve-tool-vulnogram/record.md

@@ -273,13 +273,14 @@ flow described in [*`#source` paste flow*](#source-paste-flow):
    corresponding body field on the tracker, wait for the JSON to
    regenerate, re-paste in `#source`, and re-preview.
 
-5. **Send the advisory emails.** Vulnogram dispatches to
-   `<users-list>` and `<announce-list>`.
-   **Then manually send a copy to
+5. **Send the advisory emails.** The **Send these Emails** button
+   on Vulnogram's **OSS/ASF Emails** tab dispatches, in one action,
+   to `<users-list>`, `<announce-list>`, and
    [`oss-security@lists.openwall.com`](https://oss-security.openwall.org/)
-   (required per the
-   [ASF security-committers policy](https://www.apache.org/security/committers.html)).**
-   On the tracker, add the
+   — the `oss-security` copy (required per the
+   [ASF security-committers policy](https://www.apache.org/security/committers.html))
+   is sent automatically alongside the ASF lists, so there is no
+   separate manual send. On the tracker, add the
    `announced - emails sent` label and remove `fix released`.
 
 6. **Wait for the publication-ready notification comment.** The

tools/cve-tool-vulnogram/release-manager-handoff-comment-oauth-pushed.md

@@ -123,7 +123,7 @@ With the record in `READY`, click the [OSS/ASF Emails tab](EMAIL_TAB_URL) (the `
 
 **If anything looks wrong**: don't edit it in Vulnogram. Comment on this tracker (just `@potiuk: the X field needs Y`) and we'll fix the corresponding body field here, regenerate the JSON, and re-push within the next sync. Re-preview after that.
 
-**If everything looks right**: click the **Send these Emails** button on the OSS/ASF Emails tab. The advisory ships to `USERS_LIST` and `ANNOUNCE_LIST`. **Then manually send a copy to [`oss-security@lists.openwall.com`](https://oss-security.openwall.org/) (required per the [ASF security-committers policy](https://www.apache.org/security/committers.html)).** **That is the only Vulnogram send action you make for this CVE.**
+**If everything looks right**: click the **Send these Emails** button on the OSS/ASF Emails tab. That one action ships the advisory to `USERS_LIST`, `ANNOUNCE_LIST`, and [`oss-security@lists.openwall.com`](https://oss-security.openwall.org/) — the tab is the "**OSS**/ASF Emails" tab precisely because it dispatches the `oss-security` copy (required per the [ASF security-committers policy](https://www.apache.org/security/committers.html)) automatically alongside the ASF lists. **You do not send a separate copy by hand — that is the only send action you make for this CVE.**
 
 > ⚠️ **Do not touch the tracker labels yourself.** Sync flips `fix released` → `announced - emails sent` + `announced` automatically when it sees the advisory in the public archive (usually within the same day). If you flip them manually you race the automation.
 

tools/cve-tool-vulnogram/release-manager-handoff-comment.md

@@ -125,7 +125,7 @@ With the record in `READY`, click the [OSS/ASF Emails tab](EMAIL_TAB_URL) (the `
 
 **If anything looks wrong**: don't edit it in Vulnogram. Comment on this tracker (just `@potiuk: the X field needs Y`) and we'll fix the corresponding body field here, regenerate the JSON, and re-push within the next sync. Re-preview after that.
 
-**If everything looks right**: click the **Send these Emails** button on the OSS/ASF Emails tab. The advisory ships to `USERS_LIST` and `ANNOUNCE_LIST`. **Then manually send a copy to [`oss-security@lists.openwall.com`](https://oss-security.openwall.org/) (required per the [ASF security-committers policy](https://www.apache.org/security/committers.html)).** **That is the only Vulnogram send action you make for this CVE.**
+**If everything looks right**: click the **Send these Emails** button on the OSS/ASF Emails tab. That one action ships the advisory to `USERS_LIST`, `ANNOUNCE_LIST`, and [`oss-security@lists.openwall.com`](https://oss-security.openwall.org/) — the tab is the "**OSS**/ASF Emails" tab precisely because it dispatches the `oss-security` copy (required per the [ASF security-committers policy](https://www.apache.org/security/committers.html)) automatically alongside the ASF lists. **You do not send a separate copy by hand — that is the only send action you make for this CVE.**
 
 > ⚠️ **Do not touch the tracker labels yourself.** Sync flips `fix released` → `announced - emails sent` + `announced` automatically when it sees the advisory in the public archive (usually within the same day). If you flip them manually you race the automation.