fix(sessions): make prompt repair atomic

(cherry picked from commit 55bcd8f6ccd667a5352769ad82bcbad1e73f82b1)

hermes_cli/sessions_cmd.py

@@ -1135,9 +1135,9 @@ def _cmd_repair_prompts(db, args):
 
     cleared = []
     for finding in findings:
-        db.update_system_prompt(finding["id"], None)
-        if finding["clear_pin"]:
-            db.update_session_tool_names(finding["id"], None)
+        db.clear_system_prompt_for_rebuild(
+            finding["id"], clear_tool_names=finding["clear_pin"],
+        )
         cleared.append(finding["id"])
 
     if as_json:

hermes_cli/subcommands/sessions.py

@@ -231,12 +231,12 @@ def build_sessions_parser(subparsers, *, cmd_sessions: Callable) -> None:
             "Automatic repair requires positive tools[] evidence; rows without a readable pin are "
             "reported as unverifiable and never changed by a scan. Clearing a prompt makes the next "
             "turn rebuild and persist healthy bytes. Reports without touching anything unless --apply "
-            "is given; a session_id explicitly targets one row regardless of detector evidence.")
+            "is given; a session_id is an explicit destructive override and can clear even a healthy prompt.")
     _flag(sessions_repair_prompts, "--apply", help="Clear the verified prompts (default: report only)")
     _flag(sessions_repair_prompts, "--json",
         help="Machine-readable output; with --apply, apply without an interactive confirmation")
     sessions_repair_prompts.add_argument("session_id", nargs="?", default=None,
-        help="Session id (or unique prefix) to clear explicitly, skipping detector evidence")
+        help="Destructive override: clear this session even when its stored prompt is healthy")
 
     sessions_recover = sessions_subparsers.add_parser(
         "recover", help="Rebuild canonical session data into a separate clean database",

hermes_state_sessions.py

@@ -673,6 +673,25 @@ class SessionSessionsMixin:
             self._delete_unreferenced_system_prompts(conn)
         self._execute_write(_do)
 
+    def clear_system_prompt_for_rebuild(
+        self, session_id: str, *, clear_tool_names: bool = False,
+    ) -> None:
+        """Atomically clear a stored prompt and, when requested, its tools[] pin for a fresh rebuild."""
+        def _do(conn):
+            if clear_tool_names:
+                conn.execute(
+                    "UPDATE sessions SET system_prompt_hash = NULL, system_prompt = NULL, "
+                    "tool_names = NULL WHERE id = ?",
+                    (session_id,),
+                )
+            else:
+                conn.execute(
+                    "UPDATE sessions SET system_prompt_hash = NULL, system_prompt = NULL WHERE id = ?",
+                    (session_id,),
+                )
+            self._delete_unreferenced_system_prompts(conn)
+        self._execute_write(_do)
+
     def update_session_model(
         self, session_id: str, model: str, provider: Optional[str] = None, *,
         base_url: Optional[str] = None, api_mode: Optional[str] = None,

tests/hermes_cli/test_sessions_repair_prompts.py

@@ -122,6 +122,23 @@ def test_reduced_surface_left_alone_memory_only_pin_cleared(db, monkeypatch, cap
     assert not (memory_row["tool_names"] or "")
 
 
+def test_prompt_and_memory_pin_clear_roll_back_together_on_failure(db, monkeypatch):
+    target = db.create_session("atomic-repair", "telegram", system_prompt=DEGRADED)
+    db.update_session_tool_names(target, _pin("memory"))
+
+    def _fail_gc(_conn):
+        raise RuntimeError("forced repair settlement failure")
+
+    with monkeypatch.context() as patch:
+        patch.setattr(db, "_delete_unreferenced_system_prompts", _fail_gc)
+        with pytest.raises(RuntimeError, match="forced repair settlement failure"):
+            db.clear_system_prompt_for_rebuild(target, clear_tool_names=True)
+
+    row = db.get_session(target)
+    assert row["system_prompt"] == DEGRADED
+    assert _repair_prompts_pin_names(row) == ["memory"]
+
+
 def test_positional_target_clears_memory_pin_when_prompt_is_already_null(db, monkeypatch):
     target = db.create_session("target-null", "telegram")
     db.update_session_tool_names(target, _pin("memory"))

website/docs/reference/cli-commands.md

@@ -1760,7 +1760,7 @@ Subcommands:
 | `optimize-storage` | Migrate the full-text search index to the compact v23 external-content layout; on large databases this reclaims a large fraction of `state.db`. |
 | `repair` | Repair a malformed `state.db` schema (e.g. `table messages_fts already exists`) so hidden sessions reappear; a backup is made first. |
 | `repair-routing` | Re-attach gateway conversations stranded in session rows that lost their routing identity (a chat "jumping back in time" after a restart). Dry-run by default; `--apply` performs the adoptions (stop the gateway first); `--max-gap-seconds N` tunes the contiguity window. Only unambiguous cases are repaired. See [Sessions → Repair Stranded Gateway Sessions](../user-guide/sessions.md#repair-stranded-gateway-sessions). |
-| `repair-prompts` | Report stored system prompts provably degraded by the pre-#122822 maintenance-compaction bug. Report-only by default; `--apply` clears verified rows so the next turn rebuilds them, `--json` is machine-readable (and non-interactive when combined with `--apply`), and an explicit `session_id` targets one row. Rows without a readable tools[] pin are reported as unverifiable and never auto-repaired. See [Sessions → Repair Degraded Stored Prompts](../user-guide/sessions.md#repair-degraded-stored-prompts). |
+| `repair-prompts` | Report stored system prompts provably degraded by the pre-#122822 maintenance-compaction bug. Report-only by default; `--apply` clears verified rows so the next turn rebuilds them, `--json` is machine-readable (and non-interactive when combined with `--apply`), and an explicit `session_id` is a destructive override that can clear even a healthy prompt. Rows without a readable tools[] pin are reported as unverifiable and never auto-repaired. See [Sessions → Repair Degraded Stored Prompts](../user-guide/sessions.md#repair-degraded-stored-prompts). |
 | `repair-profiles` | Settle session, routing, Telegram-topic and voice-mode state that landed under the wrong profile (rows in another profile's store, labels disagreeing with the session key, parent links crossing profiles, index rows for deleted profiles). Dry-run by default; `--apply` performs the repairs after snapshotting every store (stop the gateway first); `--legacy-main rekey\|move` decides what `agent:main` rows inside a named profile's store are; `--json` for automation. See [Sessions → Repair State Crossed Between Profiles](../user-guide/sessions.md#repair-state-crossed-between-profiles). |
 | `recover` | Offline, non-destructive recovery of a damaged `state.db` into a separate clean database. |
 | `retitle-skills` | Regenerate titles for sessions opened with a `/skill`, using what the user actually typed; lists changes unless `--apply` is passed. |

website/docs/user-guide/sessions.md

@@ -684,7 +684,8 @@ hermes sessions repair-prompts --json
 # Non-interactive automation: apply and report the ids actually cleared
 hermes sessions repair-prompts --apply --json
 
-# Explicit operator override for one session (id or unique prefix)
+# Explicit destructive override for one session (id or unique prefix).
+# This clears the stored prompt even when it is healthy.
 hermes sessions repair-prompts SESSION_ID --apply
 ```