diff --git a/.claude-plugin/marketplace.json b/.claude-plugin/marketplace.json
index b8b696ddc..f46a8eea2 100644
--- a/.claude-plugin/marketplace.json
+++ b/.claude-plugin/marketplace.json
@@ -6,14 +6,14 @@
   },
   "metadata": {
     "description": "Official Basic Memory plugins from the canonical basic-memory repository",
-    "version": "0.22.1"
+    "version": "0.23.0"
   },
   "plugins": [
     {
       "name": "basic-memory",
       "source": "./plugins/claude-code",
       "description": "The bridge between Claude's working memory and Basic Memory's durable knowledge graph \u2014 session briefings, pre-compaction checkpoints, and capture reflexes",
-      "version": "0.22.1",
+      "version": "0.23.0",
       "author": {
         "name": "Basic Machines"
       },
diff --git a/CHANGELOG.md b/CHANGELOG.md
index e69b9c2ae..7cc3ca72c 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -2,7 +2,7 @@
 
 ## Unreleased
 
-## v0.23.0 (2026-08-XX)
+## v0.23.0 (2026-08-23)
 
 Semantic search grows up and concurrent writes stop deadlocking. Search gains
 opt-in cross-encoder reranking, pluggable vector indexes with a first-party
@@ -229,6 +229,11 @@ search entries.
 - **#1073**: `edit_note` with a `memory://` URL routes to the target
   project instead of creating phantom notes (**#1066**), and scoped
   `memory://` URL paths are preserved (**#1092**).
+- **#1299**: `build_context` bounds its traversal requests — primary pages
+  cap at 50 results, related context at 100 per page, and negative
+  `max_related` is rejected instead of acting as an unbounded SQL limit —
+  so a traversal call can no longer become a multi-megabyte vault export
+  (**#1296**). Callers paginate or follow returned `memory://` links.
 - **#1285**: Glob-filtered directory listings traverse subdirectories
   again: `file_name_glob` filters results instead of accidentally pruning
   recursion, so `*.md` with depth 2 finds files inside subdirectories.
@@ -261,6 +266,15 @@ search entries.
   `create_time` — timestamps fall back to `update_time`, then the earliest
   message time, so the whole archive imports (**#1276**); undecodable
   import uploads return a 400 with the parse error instead of a 500.
+- **#1298**: The update check no longer reports "up to date" when the
+  Homebrew probe fails, no longer auto-runs `brew upgrade` for an update
+  inferred from PyPI metadata, and names the actual Homebrew target
+  version it will install.
+- **#1300 / #1302**: The multilingual-E5 prefix contract is documented —
+  asymmetric FastEmbed models need `semantic_embedding_query_prefix` /
+  `semantic_embedding_document_prefix`, which already drive reindex through
+  the embedding identity — along with custom-model selection, the installed
+  FastEmbed catalog, and dimensions (**#1264**).
 - **#1058 / #1094**: `bm doctor` never prints a blank failure message, and
   migrations adapt to existing event loops (**#1027**).
 - **#1080**: Loading config no longer recreates an empty `~/basic-memory`
@@ -279,7 +293,9 @@ search entries.
   registration APIs instead of private PluginManager writes, preserving the
   lifecycle ownership that cleans them up on provider unload (**#1257**);
   the slash-command monkeypatch docs are split by Hermes version so modern
-  installs skip the legacy collector workaround (**#1278**).
+  installs skip the legacy collector workaround (**#1278**), and the install
+  docs require a Hermes release with managed manifest-v2 installation
+  instead of the nonexistent `--path` option (**#1280**, **#1303**).
 
 ### Maintenance
 
@@ -292,6 +308,12 @@ search entries.
 - Kept Milvus as a first-party optional vector backend while removing the
   unused Python entry-point registry for separately packaged vector
   adapters.
+- **#1292 / #1293**: The dev-release pipeline publishes again via PyPI
+  trusted publishing — its version gate imported a hardcoded module version
+  and had silently skipped every dev publish since the 0.22.1 bump.
+- **#1304 / #1307**: The concurrent-write convergence benchmark is ported
+  into the canonical `/benchmarks` package, and the read-load benchmark
+  runs against the current MCP SDK's typed result fields.
 
 ## v0.22.1 (2026-06-12)
 
diff --git a/benchmarks/AGENTS.md b/benchmarks/AGENTS.md
index 7827cc1be..d94b2ccdd 100644
--- a/benchmarks/AGENTS.md
+++ b/benchmarks/AGENTS.md
@@ -7,7 +7,7 @@ directory for comparing Basic Memory against other memory systems.
 
 Primary goals:
 - Deterministic retrieval benchmarks
-- Optional LLM-as-a-judge benchmarks
+- End-to-end QA scoring with fixed answerer and judge models
 - Public, reproducible artifact publication (including provenance metadata)
 
 The benchmark package keeps its own `pyproject.toml` and lockfile so benchmark
@@ -16,7 +16,6 @@ dependencies do not pollute the Core product environment.
 ## Build / Test Commands
 
 - Install: `uv sync --group dev`
-- Install judge extras: `uv sync --group dev --extra judge`
 - Run tests: `uv run pytest -q`
 - Lint: `uv run ruff check .`
 - Type check: `uv run pyright`
@@ -37,8 +36,8 @@ Dataset and conversion:
 Run retrieval:
 - `uv run bm-bench run retrieval --providers bm-local,mem0-local --dataset-id locomo --dataset-path benchmarks/datasets/locomo/locomo10.json --corpus-dir benchmarks/generated/locomo/docs --queries-path benchmarks/generated/locomo/queries.json --output-root benchmarks/runs --allow-provider-skip`
 
-Run judge (optional):
-- `uv run bm-bench run judge --run-dir benchmarks/runs/<run-id>`
+Run end-to-end QA scoring:
+- `uv run bm-bench run qa --run-dir benchmarks/runs/<run-id> --answerer claude:claude-haiku-4-5 --judge claude:claude-sonnet-4-6`
 
 Validate and publish:
 - `uv run bm-bench validate-artifacts --run-dir benchmarks/runs/<run-id>`
@@ -53,15 +52,15 @@ Validate and publish:
 - `just bench-run-bm-local`
 - `just bench-run-mem0-local`
 - `just bench-run-full`
-- `just bench-judge RUN_DIR=benchmarks/runs/<run-id>`
-- `just bench-publish RUN_DIR=benchmarks/runs/<run-id>`
+- `just bench-qa benchmarks/runs/<run-id>`
+- `just bench-publish benchmarks/runs/<run-id>`
 
 ## Repository Layout
 
 - `src/basic_memory_benchmarks/cli.py` - CLI surface
 - `src/basic_memory_benchmarks/runner.py` - run orchestration
 - `src/basic_memory_benchmarks/providers/` - provider adapters (`bm-local`, `bm-cloud`, `mem0-local`, `zep-reference`)
-- `src/basic_memory_benchmarks/scoring/` - retrieval + judge scoring
+- `src/basic_memory_benchmarks/scoring/` - retrieval + end-to-end QA scoring
 - `src/basic_memory_benchmarks/reporting/` - artifact writers / comparison helpers
 - `src/basic_memory_benchmarks/converters/` - dataset conversion logic
 - `src/basic_memory_benchmarks/datasets/` - dataset fetch/load helpers
diff --git a/benchmarks/README.md b/benchmarks/README.md
index 9b67ba830..b47cf9275 100644
--- a/benchmarks/README.md
+++ b/benchmarks/README.md
@@ -7,7 +7,7 @@ dependencies remain isolated from the product environment.
 ## Goals
 
 - Deterministic retrieval benchmarks (Recall@5/10, MRR, Precision@5, content-hit, latency)
-- Optional LLM-as-judge scoring (Pydantic Evals)
+- End-to-end QA scoring with fixed answerer and judge models
 - Public artifacts with provenance and reproducibility metadata
 - Clean dependency isolation from the core Basic Memory package
 
@@ -29,12 +29,6 @@ dependencies remain isolated from the product environment.
 uv sync --group dev
 ```
 
-Optional judge dependencies:
-
-```bash
-uv sync --group dev --extra judge
-```
-
 ## Quickstart
 
 ### 1) Fetch LoCoMo dataset
@@ -85,16 +79,7 @@ know") when the retrieved memories don't contain the answer, and abstention is
 graded correct only when the gold answer marks the question unanswerable
 (LoCoMo adversarial cases).
 
-### 5) Optional retrieval-context judge (legacy)
-
-Scores whether the *retrieved context* contains the expected answer, without
-answer generation:
-
-```bash
-uv run bm-bench run judge --run-dir benchmarks/runs/<run-id>
-```
-
-### 6) Publish run artifacts
+### 5) Publish run artifacts
 
 ```bash
 uv run bm-bench publish --run-dir benchmarks/runs/<run-id>
@@ -290,8 +275,6 @@ Per run (`benchmarks/runs/<run-id>/`):
 - `retrieval-summary.json`
 - `per-query-qa.jsonl` (optional)
 - `qa-summary.json` (optional)
-- `per-query-judge.jsonl` (optional)
-- `judge-summary.json` (optional)
 - `summary.md`
 
 ## Just commands
@@ -303,8 +286,8 @@ just bench-convert-locomo
 just bench-run-bm-local
 just bench-run-mem0-local
 just bench-run-full
-just bench-judge
-just bench-publish RUN_DIR=benchmarks/runs/<run-id>
+just bench-qa benchmarks/runs/<run-id>
+just bench-publish benchmarks/runs/<run-id>
 ```
 
 ## Notes on dataset publication
diff --git a/benchmarks/docs/benchmarks.md b/benchmarks/docs/benchmarks.md
index 17952f07f..0fd0911ce 100644
--- a/benchmarks/docs/benchmarks.md
+++ b/benchmarks/docs/benchmarks.md
@@ -12,9 +12,9 @@ It covers:
 
 | Area | Status |
 | --- | --- |
-| Single run execution (`run retrieval`, `run full`, `run judge`) | Implemented |
+| Single run execution (`run retrieval`, `run full`, `run qa`) | Implemented |
 | Concurrent write convergence (`run concurrent-write`) | Implemented |
-| `just` one-command pipelines (`bench-full`, `bench-full-judge`) | Implemented |
+| `just` retrieval and QA workflows (`bench-full`, `bench-qa`) | Implemented |
 | Artifact generation and publish/compare commands | Implemented |
 | Manual BM revision comparison via worktrees + `--bm-local-path` | Implemented workflow, manual orchestration |
 | `bm-bench run revision-matrix` | Planned, not implemented yet |
@@ -58,12 +58,6 @@ cd /path/to/basic-memory/benchmarks
 just sync
 ```
 
-If you plan to run judge metrics:
-
-```bash
-just sync-judge
-```
-
 ### Dataset assumptions
 
 LoCoMo source and converted outputs are created by:
@@ -78,7 +72,7 @@ just bench-prepare-long
 ### `just` commands (current)
 
 - `bench-full`
-- `bench-full-judge`
+- `bench-qa`
 - `bench-concurrent-write-smoke`
 - `bench-concurrent-write-load`
 - `bench-prepare-short`
@@ -86,7 +80,6 @@ just bench-prepare-long
 - `bench-run-short`
 - `bench-run-long`
 - `bench-run-full`
-- `bench-judge`
 - `bench-validate`
 - `bench-publish`
 - `bench-compare`
@@ -101,7 +94,9 @@ Top-level commands:
 - `run retrieval`
 - `run concurrent-write`
 - `run full`
-- `run judge`
+- `run qa`
+- `run rejudge`
+- `run review`
 - `compare`
 - `validate-artifacts`
 - `publish`
@@ -120,17 +115,15 @@ This runs:
 2. `just bench-prepare-long`
 3. `just bench-run-full`
 
-### One-command full retrieval + judge
+### End-to-end QA scoring
 
 ```bash
 cd /path/to/basic-memory/benchmarks
-just bench-full-judge
+just bench-qa benchmarks/runs/<run_id>
 ```
 
-This runs:
-1. `just sync-judge`
-2. `just bench-prepare-long`
-3. `just bench-run-full-judge`
+This generates answers from each provider's retrieved context, applies the
+same judge to every provider, and writes QA artifacts into the retrieval run.
 
 ### Short vs long workflows
 
@@ -196,10 +189,15 @@ Required files:
 - `retrieval-summary.json`
 - `summary.md`
 
-Optional judge files:
+Optional QA files:
 
-- `per-query-judge.jsonl`
-- `judge-summary.json`
+- `per-query-qa.jsonl`
+- `qa-summary.json`
+- `per-query-qa-rejudge.jsonl`
+- `qa-rejudge-summary.json`
+- `qa-rejudge-flips.json`
+- `review.html`
+- `qa-diagnosis.json`
 
 ### Key provenance fields
 
@@ -483,7 +481,7 @@ Dry-run checks:
 just --dry-run bench-run-short
 just --dry-run bench-run-long
 just --dry-run bench-full
-just --dry-run bench-full-judge
+just --dry-run bench-qa benchmarks/runs/<run_id>
 ```
 
 Artifact field checks:
diff --git a/benchmarks/justfile b/benchmarks/justfile
index 6bcff0ae9..eaef7ce99 100644
--- a/benchmarks/justfile
+++ b/benchmarks/justfile
@@ -18,9 +18,6 @@ longmemeval_dev_output_dir := "benchmarks/generated/longmemeval-s-dev"
 sync:
     uv sync --group dev
 
-sync-judge:
-    uv sync --group dev --extra judge
-
 test:
     uv run pytest -q
 
@@ -110,13 +107,6 @@ bench-full:
     just bench-prepare-long
     just bench-run-full
 
-# Full retrieval + judge pipeline:
-# 1) sync deps (+judge extras), 2) fetch+convert long dataset, 3) run full with judge
-bench-full-judge model="gpt-4o-mini":
-    just sync-judge
-    just bench-prepare-long
-    just bench-run-full-judge model="{{model}}"
-
 # --- Benchmark execution ---
 
 bench-smoke:
@@ -198,18 +188,6 @@ bench-run-full:
       {{bm_local_path_flag}} \
       --allow-provider-skip
 
-bench-run-full-judge model="gpt-4o-mini":
-    uv run bm-bench run full \
-      --dataset-id locomo \
-      --dataset-path {{locomo_dataset_path}} \
-      --corpus-dir benchmarks/generated/locomo/docs \
-      --queries-path benchmarks/generated/locomo/queries.json \
-      --providers bm-local,mem0-local \
-      {{bm_local_path_flag}} \
-      --allow-provider-skip \
-      --judge \
-      --judge-model "{{model}}"
-
 # --- Concurrency benchmark (basic-memory#1248) ---
 
 # Small-scale smoke: 4 writers x 25 notes; strict so divergence fails the command
@@ -244,8 +222,11 @@ bench-latest-run:
     set -euo pipefail
     ls -1dt benchmarks/runs/* | head -n 1
 
-bench-judge run_dir model="gpt-4o-mini":
-    uv run bm-bench run judge --run-dir "{{run_dir}}" --model "{{model}}"
+bench-qa run_dir answerer="claude:claude-haiku-4-5" judge="claude:claude-sonnet-4-6":
+    uv run bm-bench run qa \
+      --run-dir "{{run_dir}}" \
+      --answerer "{{answerer}}" \
+      --judge "{{judge}}"
 
 bench-validate run_dir:
     uv run bm-bench validate-artifacts --run-dir "{{run_dir}}"
diff --git a/benchmarks/pyproject.toml b/benchmarks/pyproject.toml
index f206ef66e..d858f0792 100644
--- a/benchmarks/pyproject.toml
+++ b/benchmarks/pyproject.toml
@@ -18,11 +18,6 @@ dependencies = [
     "typer>=0.16.1",
 ]
 
-[project.optional-dependencies]
-judge = [
-    "pydantic-evals>=0.4.0",
-]
-
 [project.scripts]
 bm-bench = "basic_memory_benchmarks.cli:main"
 basic-memory-benchmarks = "basic_memory_benchmarks.cli:main"
diff --git a/benchmarks/scripts/read_load_bench.py b/benchmarks/scripts/read_load_bench.py
index ad2465e7d..35bb2de45 100644
--- a/benchmarks/scripts/read_load_bench.py
+++ b/benchmarks/scripts/read_load_bench.py
@@ -382,7 +382,7 @@ def write_manifest(
 
 
 def result_payload(result: CallToolResult) -> dict[str, object]:
-    structured = result.structuredContent
+    structured = result.structured_content
     if isinstance(structured, dict):
         wrapped = structured.get("result")
         payload = wrapped if isinstance(wrapped, dict) else structured
@@ -402,7 +402,7 @@ def result_payload(result: CallToolResult) -> dict[str, object]:
 
 def result_text(result: CallToolResult) -> str | None:
     """Return text from a successful text-mode tool response."""
-    if result.isError:
+    if result.is_error:
         return None
     text_parts = [
         text for item in result.content if isinstance((text := getattr(item, "text", None)), str)
@@ -517,7 +517,7 @@ async def write_target(
             "output_format": "json",
         },
     )
-    if result.isError:
+    if result.is_error:
         raise RuntimeError(f"write_note failed while seeding {title}")
     payload = result_payload(result)
     identifier = payload.get("permalink")
@@ -573,7 +573,7 @@ async def searchable_count(session: ClientSession, project: str) -> int:
             "output_format": "json",
         },
     )
-    if result.isError:
+    if result.is_error:
         return 0
     payload = result_payload(result)
     total = payload.get("total")
@@ -742,7 +742,7 @@ async def run(args: argparse.Namespace) -> int:
                 "create_memory_project",
                 {"project_name": project, "project_path": str(project_dir)},
             )
-            if created.isError:
+            if created.is_error:
                 raise RuntimeError("could not create benchmark project")
 
             targets = await seed_corpus(
diff --git a/benchmarks/src/basic_memory_benchmarks/cli.py b/benchmarks/src/basic_memory_benchmarks/cli.py
index 42d33cc12..8f2739738 100644
--- a/benchmarks/src/basic_memory_benchmarks/cli.py
+++ b/benchmarks/src/basic_memory_benchmarks/cli.py
@@ -30,7 +30,6 @@
 )
 from basic_memory_benchmarks.runner import (
     run_diagnose_stage,
-    run_judge,
     run_qa_stage,
     run_rejudge_stage,
     run_review_stage,
@@ -430,15 +429,6 @@ def run_rejudge_command(
     console.print(f"Flips: [cyan]{out / 'qa-rejudge-flips.json'}[/cyan]")
 
 
-@run_app.command("judge")
-def run_judge_command(
-    run_dir: Path = typer.Option(..., "--run-dir"),
-    model: str = typer.Option("gpt-4o-mini", "--model"),
-) -> None:
-    out = run_judge(run_dir=run_dir, model=model)
-    console.print(f"Judge run complete: [green]{out}[/green]")
-
-
 @run_app.command("full")
 def run_full_command(
     providers: str = typer.Option("bm-local,mem0-local", "--providers"),
@@ -456,8 +446,6 @@ def run_full_command(
     bm_source: str = typer.Option("github:basicmachines-co/basic-memory@main", "--bm-source"),
     bm_local_path: str | None = typer.Option(None, "--bm-local-path"),
     allow_provider_skip: bool = typer.Option(True, "--allow-provider-skip/--strict-providers"),
-    judge: bool = typer.Option(False, "--judge"),
-    judge_model: str = typer.Option("gpt-4o-mini", "--judge-model"),
 ) -> None:
     run_retrieval_command(
         providers=providers,
@@ -473,18 +461,6 @@ def run_full_command(
         allow_provider_skip=allow_provider_skip,
     )
 
-    if judge:
-        resolved_run_id = run_id
-        if resolved_run_id is None:
-            # run_retrieval_command generated uuid when run_id is None. infer by latest dir.
-            run_dirs = sorted(Path(output_root).glob("*"), key=lambda path: path.stat().st_mtime)
-            if not run_dirs:
-                raise RuntimeError("Unable to locate run directory for judge step")
-            run_dir = run_dirs[-1]
-        else:
-            run_dir = Path(output_root) / resolved_run_id
-        run_judge_command(run_dir=run_dir, model=judge_model)
-
 
 @app.command("compare")
 def compare_runs(
diff --git a/benchmarks/src/basic_memory_benchmarks/llm/runners.py b/benchmarks/src/basic_memory_benchmarks/llm/runners.py
index 2cf72915e..a328cc9c5 100644
--- a/benchmarks/src/basic_memory_benchmarks/llm/runners.py
+++ b/benchmarks/src/basic_memory_benchmarks/llm/runners.py
@@ -16,6 +16,7 @@
 from __future__ import annotations
 
 import json
+import os
 import subprocess
 import time
 from abc import ABC, abstractmethod
@@ -187,7 +188,8 @@ def create_runner(spec: str, *, api_key: str | None = None) -> LLMRunner:
             raise ValueError(
                 f"openai-compat spec must be 'openai-compat:<model>@<base_url>', got: {spec}"
             )
-        return OpenAICompatRunner(model=model, base_url=base_url, api_key=api_key)
+        resolved_api_key = api_key if api_key is not None else os.getenv("OPENAI_API_KEY")
+        return OpenAICompatRunner(model=model, base_url=base_url, api_key=resolved_api_key)
     raise ValueError(
         f"Unknown runner spec '{spec}'. Expected 'claude:<model>' or "
         f"'openai-compat:<model>@<base_url>'."
diff --git a/benchmarks/src/basic_memory_benchmarks/models.py b/benchmarks/src/basic_memory_benchmarks/models.py
index a7d88c4aa..d5fd71772 100644
--- a/benchmarks/src/basic_memory_benchmarks/models.py
+++ b/benchmarks/src/basic_memory_benchmarks/models.py
@@ -79,26 +79,6 @@ class RetrievalSummary(BaseModel):
     adversarial_breakout: RetrievalMetrics
 
 
-class JudgeCaseResult(BaseModel):
-    provider: str
-    query_id: str
-    category: str
-    passed: bool
-    score: float
-    reason: str
-    evaluator: str
-
-
-class JudgeSummary(BaseModel):
-    provider: str
-    evaluator: str
-    model: str
-    total_cases: int
-    pass_count: int
-    accuracy: float
-    skipped_reason: str | None = None
-
-
 class QACategoryMetrics(BaseModel):
     total: int = 0
     correct: int = 0
@@ -213,8 +193,6 @@ class RunConfig(BaseModel):
     top_k: int = 10
     bm_source: str = "github:basicmachines-co/basic-memory@main"
     bm_local_path: str | None = None
-    judge_enabled: bool = False
-    judge_model: str = "gpt-4o-mini"
     allow_provider_skip: bool = True
 
 
@@ -237,8 +215,6 @@ class RunArtifacts(BaseModel):
     provider_status: list[ProviderStatus]
     retrieval_summaries: list[RetrievalSummary]
     retrieval_rows: list[PerQueryRetrievalResult]
-    judge_summaries: list[JudgeSummary] = Field(default_factory=list)
-    judge_rows: list[JudgeCaseResult] = Field(default_factory=list)
     fairness_warnings: list[str] = Field(default_factory=list)
 
 
diff --git a/benchmarks/src/basic_memory_benchmarks/reporting/artifacts.py b/benchmarks/src/basic_memory_benchmarks/reporting/artifacts.py
index 6ca49d22c..b673422fa 100644
--- a/benchmarks/src/basic_memory_benchmarks/reporting/artifacts.py
+++ b/benchmarks/src/basic_memory_benchmarks/reporting/artifacts.py
@@ -6,8 +6,6 @@
 from pathlib import Path
 
 from basic_memory_benchmarks.models import (
-    JudgeCaseResult,
-    JudgeSummary,
     PerQueryRetrievalResult,
     ProviderStatus,
     RetrievalSummary,
@@ -35,8 +33,6 @@ def write_artifacts(
     retrieval_rows: list[PerQueryRetrievalResult],
     retrieval_summaries: list[RetrievalSummary],
     fairness_warnings: list[str],
-    judge_rows: list[JudgeCaseResult] | None = None,
-    judge_summaries: list[JudgeSummary] | None = None,
 ) -> None:
     run_dir.mkdir(parents=True, exist_ok=True)
     _write_json(run_dir / "manifest.json", manifest.model_dump(mode="json"))
@@ -56,25 +52,11 @@ def write_artifacts(
         },
     )
 
-    if judge_rows is not None:
-        _write_jsonl(
-            run_dir / "per-query-judge.jsonl",
-            [row.model_dump(mode="json") for row in judge_rows],
-        )
-    if judge_summaries is not None:
-        _write_json(
-            run_dir / "judge-summary.json",
-            {
-                "providers": [row.model_dump(mode="json") for row in judge_summaries],
-            },
-        )
-
     summary_markdown = build_summary_markdown(
         manifest=manifest,
         provider_status=provider_status,
         retrieval_summaries=retrieval_summaries,
         fairness_warnings=fairness_warnings,
-        judge_summaries=judge_summaries or [],
     )
     (run_dir / "summary.md").write_text(summary_markdown, encoding="utf-8")
 
@@ -85,7 +67,6 @@ def build_summary_markdown(
     provider_status: list[ProviderStatus],
     retrieval_summaries: list[RetrievalSummary],
     fairness_warnings: list[str],
-    judge_summaries: list[JudgeSummary],
 ) -> str:
     lines: list[str] = []
     lines.append(f"# Benchmark Run `{manifest.run_id}`")
@@ -146,19 +127,6 @@ def build_summary_markdown(
         )
     lines.append("")
 
-    if judge_summaries:
-        lines.append("## Judge Summary")
-        lines.append("")
-        lines.append("| Provider | Evaluator | Model | Cases | Accuracy | Note |")
-        lines.append("| --- | --- | --- | --- | --- | --- |")
-        for summary in judge_summaries:
-            lines.append(
-                "| "
-                f"{summary.provider} | {summary.evaluator} | {summary.model} | "
-                f"{summary.total_cases} | {summary.accuracy:.3f} | {summary.skipped_reason or ''} |"
-            )
-        lines.append("")
-
     if fairness_warnings:
         lines.append("## Fairness Warnings")
         lines.append("")
diff --git a/benchmarks/src/basic_memory_benchmarks/runner.py b/benchmarks/src/basic_memory_benchmarks/runner.py
index 2a4f63622..d769a1f06 100644
--- a/benchmarks/src/basic_memory_benchmarks/runner.py
+++ b/benchmarks/src/basic_memory_benchmarks/runner.py
@@ -22,7 +22,6 @@
 from basic_memory_benchmarks.providers import create_provider
 from basic_memory_benchmarks.providers.base import BenchmarkProvider
 from basic_memory_benchmarks.reporting.artifacts import write_artifacts
-from basic_memory_benchmarks.scoring.judge import run_optional_judge
 from basic_memory_benchmarks.scoring.retrieval import evaluate_query, summarize_provider
 from basic_memory_benchmarks.utils import (
     git_sha,
@@ -487,50 +486,3 @@ def run_rejudge_stage(
         encoding="utf-8",
     )
     return run_dir
-
-
-def run_judge(
-    *,
-    run_dir: Path,
-    model: str,
-) -> Path:
-    retrieval_path = run_dir / "per-query-retrieval.jsonl"
-    if not retrieval_path.exists():
-        raise FileNotFoundError(f"Missing retrieval artifact: {retrieval_path}")
-
-    rows: list[PerQueryRetrievalResult] = []
-    with retrieval_path.open("r", encoding="utf-8") as file:
-        for line in file:
-            line = line.strip()
-            if not line:
-                continue
-            rows.append(PerQueryRetrievalResult.model_validate(json.loads(line)))
-
-    grouped: dict[str, list[PerQueryRetrievalResult]] = {}
-    for row in rows:
-        grouped.setdefault(row.provider, []).append(row)
-
-    judge_rows = []
-    judge_summaries = []
-    for provider, provider_rows in grouped.items():
-        provider_case_results, provider_summary = run_optional_judge(
-            provider_rows,
-            provider=provider,
-            model=model,
-        )
-        judge_rows.extend(provider_case_results)
-        judge_summaries.append(provider_summary)
-
-    judge_jsonl = run_dir / "per-query-judge.jsonl"
-    with judge_jsonl.open("w", encoding="utf-8") as file:
-        for row in judge_rows:
-            file.write(json.dumps(row.model_dump(mode="json"), sort_keys=True) + "\n")
-
-    judge_summary_path = run_dir / "judge-summary.json"
-    judge_summary_path.write_text(
-        json.dumps(
-            {"providers": [item.model_dump(mode="json") for item in judge_summaries]}, indent=2
-        ),
-        encoding="utf-8",
-    )
-    return run_dir
diff --git a/benchmarks/src/basic_memory_benchmarks/scoring/judge.py b/benchmarks/src/basic_memory_benchmarks/scoring/judge.py
deleted file mode 100644
index 755205559..000000000
--- a/benchmarks/src/basic_memory_benchmarks/scoring/judge.py
+++ /dev/null
@@ -1,140 +0,0 @@
-"""Optional judge scoring.
-
-Primary mode tries Pydantic Evals. If unavailable or misconfigured, it falls back to
-an explicit deterministic checker and records the reason.
-"""
-
-from __future__ import annotations
-
-import os
-from typing import Any
-
-from basic_memory_benchmarks.models import JudgeCaseResult, JudgeSummary, PerQueryRetrievalResult
-
-
-def _deterministic_contains_eval(
-    rows: list[PerQueryRetrievalResult], provider: str
-) -> tuple[list[JudgeCaseResult], JudgeSummary]:
-    case_results: list[JudgeCaseResult] = []
-    for row in rows:
-        expected = (row.expected_answer or "").strip().lower()
-        context = row.retrieved_context.strip().lower()
-        passed = bool(expected and expected in context)
-        case_results.append(
-            JudgeCaseResult(
-                provider=provider,
-                query_id=row.query_id,
-                category=row.category,
-                passed=passed,
-                score=1.0 if passed else 0.0,
-                reason="Deterministic contains check",
-                evaluator="deterministic-fallback",
-            )
-        )
-
-    pass_count = sum(1 for item in case_results if item.passed)
-    total = len(case_results)
-    summary = JudgeSummary(
-        provider=provider,
-        evaluator="deterministic-fallback",
-        model="none",
-        total_cases=total,
-        pass_count=pass_count,
-        accuracy=(pass_count / total) if total else 0.0,
-    )
-    return case_results, summary
-
-
-def run_optional_judge(
-    rows: list[PerQueryRetrievalResult],
-    provider: str,
-    model: str,
-) -> tuple[list[JudgeCaseResult], JudgeSummary]:
-    """Run optional judge scoring.
-
-    If pydantic-evals + OpenAI credentials are available, we attempt to use it.
-    Otherwise we return deterministic fallback scores.
-    """
-    relevant_rows = [row for row in rows if row.expected_answer]
-    if not relevant_rows:
-        return [], JudgeSummary(
-            provider=provider,
-            evaluator="none",
-            model=model,
-            total_cases=0,
-            pass_count=0,
-            accuracy=0.0,
-            skipped_reason="No expected answers available for judge scoring",
-        )
-
-    if not os.getenv("OPENAI_API_KEY"):
-        case_results, summary = _deterministic_contains_eval(relevant_rows, provider)
-        summary.skipped_reason = "OPENAI_API_KEY missing; used deterministic fallback"
-        return case_results, summary
-
-    try:
-        from pydantic_evals import Case, Dataset  # type: ignore
-        from pydantic_evals.evaluators import LLMJudge  # type: ignore
-    except Exception:
-        case_results, summary = _deterministic_contains_eval(relevant_rows, provider)
-        summary.skipped_reason = "pydantic-evals not installed; used deterministic fallback"
-        return case_results, summary
-
-    # Best-effort pydantic-evals integration.
-    # Trigger: runtime has judge deps + API key
-    # Why: align with competitor methodology
-    # Outcome: if API shape changes, fallback remains deterministic and explicit
-    try:
-        cases: list[Any] = []
-        for row in relevant_rows:
-            rubric = (
-                "PASS if the candidate response contains the same core factual answer as expected. "
-                "FAIL if key facts are missing or contradictory."
-            )
-            candidate = row.retrieved_context
-            expected = row.expected_answer or ""
-            cases.append(
-                Case(
-                    name=row.query_id,
-                    inputs=f"question: {row.query_text}\nexpected: {expected}\ncandidate: {candidate}",
-                    evaluators=[LLMJudge(rubric=rubric, include_input=True, model=model)],
-                    metadata={"provider": provider, "category": row.category},
-                )
-            )
-
-        dataset = Dataset(cases=cases)
-        report = dataset.evaluate_sync(lambda value: value)
-
-        case_results: list[JudgeCaseResult] = []
-        pass_count = 0
-        for idx, row in enumerate(relevant_rows):
-            case_report = report.cases[idx]
-            passed = bool(getattr(case_report, "passed", False))
-            pass_count += 1 if passed else 0
-            reason = str(getattr(case_report, "reason", "")) or "LLM judge result"
-            case_results.append(
-                JudgeCaseResult(
-                    provider=provider,
-                    query_id=row.query_id,
-                    category=row.category,
-                    passed=passed,
-                    score=1.0 if passed else 0.0,
-                    reason=reason,
-                    evaluator="pydantic-evals-llmjudge",
-                )
-            )
-
-        total = len(case_results)
-        summary = JudgeSummary(
-            provider=provider,
-            evaluator="pydantic-evals-llmjudge",
-            model=model,
-            total_cases=total,
-            pass_count=pass_count,
-            accuracy=(pass_count / total) if total else 0.0,
-        )
-        return case_results, summary
-    except Exception:
-        case_results, summary = _deterministic_contains_eval(relevant_rows, provider)
-        summary.skipped_reason = "pydantic-evals execution failed; used deterministic fallback"
-        return case_results, summary
diff --git a/benchmarks/tests/llm/test_runners.py b/benchmarks/tests/llm/test_runners.py
index 55d74a829..d1a29685d 100644
--- a/benchmarks/tests/llm/test_runners.py
+++ b/benchmarks/tests/llm/test_runners.py
@@ -28,6 +28,14 @@ def test_openai_compat_spec(self):
         assert runner.model == "llama3.1"
         assert runner.base_url == "http://localhost:11434/v1"
 
+    def test_openai_compat_uses_openai_api_key(self, monkeypatch):
+        monkeypatch.setenv("OPENAI_API_KEY", "test-key")
+
+        runner = create_runner("openai-compat:gpt-4o-mini@https://api.openai.com/v1")
+
+        assert isinstance(runner, OpenAICompatRunner)
+        assert runner._api_key == "test-key"
+
     def test_openai_compat_spec_requires_base_url(self):
         with pytest.raises(ValueError):
             create_runner("openai-compat:llama3.1")
diff --git a/benchmarks/tests/test_cli_surface.py b/benchmarks/tests/test_cli_surface.py
new file mode 100644
index 000000000..52fb92709
--- /dev/null
+++ b/benchmarks/tests/test_cli_surface.py
@@ -0,0 +1,24 @@
+from typer.testing import CliRunner
+
+from basic_memory_benchmarks.cli import app
+
+
+runner = CliRunner()
+
+
+def test_modern_qa_replaces_legacy_judge_command() -> None:
+    qa_help = runner.invoke(app, ["run", "qa", "--help"])
+    legacy_judge = runner.invoke(app, ["run", "judge", "--help"])
+
+    assert qa_help.exit_code == 0
+    assert "--answerer" in qa_help.output
+    assert "--judge" in qa_help.output
+    assert legacy_judge.exit_code != 0
+    assert "No such command 'judge'" in legacy_judge.output
+
+
+def test_full_command_has_no_legacy_judge_options() -> None:
+    result = runner.invoke(app, ["run", "full", "--help"])
+
+    assert result.exit_code == 0
+    assert "--judge-model" not in result.output
diff --git a/benchmarks/tests/test_manifest_schema.py b/benchmarks/tests/test_manifest_schema.py
index 0c16441bd..88d5d3a1c 100644
--- a/benchmarks/tests/test_manifest_schema.py
+++ b/benchmarks/tests/test_manifest_schema.py
@@ -30,3 +30,5 @@ def test_manifest_schema_roundtrip() -> None:
     payload = manifest.model_dump(mode="json")
     assert payload["run_id"] == "run1"
     assert payload["config"]["providers"] == ["bm-local"]
+    assert "judge_enabled" not in payload["config"]
+    assert "judge_model" not in payload["config"]
diff --git a/benchmarks/uv.lock b/benchmarks/uv.lock
index 4948a715f..da9f5c6eb 100644
--- a/benchmarks/uv.lock
+++ b/benchmarks/uv.lock
@@ -216,11 +216,6 @@ dependencies = [
     { name = "typer" },
 ]
 
-[package.optional-dependencies]
-judge = [
-    { name = "pydantic-evals" },
-]
-
 [package.dev-dependencies]
 dev = [
     { name = "pyright" },
@@ -236,12 +231,10 @@ requires-dist = [
     { name = "mcp", specifier = ">=1.23.1" },
     { name = "mem0ai", extras = ["nlp"], specifier = "==2.0.5" },
     { name = "pydantic", specifier = ">=2.12.0" },
-    { name = "pydantic-evals", marker = "extra == 'judge'", specifier = ">=0.4.0" },
     { name = "python-frontmatter", specifier = ">=1.1.0" },
     { name = "rich", specifier = ">=13.9.4" },
     { name = "typer", specifier = ">=0.16.1" },
 ]
-provides-extras = ["judge"]
 
 [package.metadata.requires-dev]
 dev = [
@@ -898,19 +891,6 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/e6/ab/fb21f4c939bb440104cc2b396d3be1d9b7a9fd3c6c2a53d98c45b3d7c954/fsspec-2026.2.0-py3-none-any.whl", hash = "sha256:98de475b5cb3bd66bedd5c4679e87b4fdfe1a3bf4d707b151b3c07e58c9a2437", size = 202505, upload-time = "2026-02-05T21:50:51.819Z" },
 ]
 
-[[package]]
-name = "genai-prices"
-version = "0.0.54"
-source = { registry = "https://pypi.org/simple" }
-dependencies = [
-    { name = "httpx" },
-    { name = "pydantic" },
-]
-sdist = { url = "https://files.pythonhosted.org/packages/8e/20/d64d04c7870db01232f8f8e36dec2072281b1f594b924200aa017778eec2/genai_prices-0.0.54.tar.gz", hash = "sha256:9d985affc19d055be16613b0c5e49d182d4c2164cf1587129f9996dfb9a3cb8d", size = 59588, upload-time = "2026-02-17T20:26:06.849Z" }
-wheels = [
-    { url = "https://files.pythonhosted.org/packages/34/71/d2a941b0ca01186912fedb096d8eef7b3e1680c86fdcf8fe3dc84e76d5a9/genai_prices-0.0.54-py3-none-any.whl", hash = "sha256:5b45012b2981b7d4d42c49c8614ee95420fec244c87542542045786b36fc2235", size = 62198, upload-time = "2026-02-17T20:26:05.186Z" },
-]
-
 [[package]]
 name = "greenlet"
 version = "3.3.2"
@@ -954,15 +934,6 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/29/4b/45d90626aef8e65336bed690106d1382f7a43665e2249017e9527df8823b/greenlet-3.3.2-cp314-cp314t-win_amd64.whl", hash = "sha256:c04c5e06ec3e022cbfe2cd4a846e1d4e50087444f875ff6d2c2ad8445495cf1a", size = 237086, upload-time = "2026-02-20T20:20:45.786Z" },
 ]
 
-[[package]]
-name = "griffelib"
-version = "2.0.0"
-source = { registry = "https://pypi.org/simple" }
-sdist = { url = "https://files.pythonhosted.org/packages/ad/06/eccbd311c9e2b3ca45dbc063b93134c57a1ccc7607c5e545264ad092c4a9/griffelib-2.0.0.tar.gz", hash = "sha256:e504d637a089f5cab9b5daf18f7645970509bf4f53eda8d79ed71cce8bd97934", size = 166312, upload-time = "2026-03-23T21:06:55.954Z" }
-wheels = [
-    { url = "https://files.pythonhosted.org/packages/4d/51/c936033e16d12b627ea334aaaaf42229c37620d0f15593456ab69ab48161/griffelib-2.0.0-py3-none-any.whl", hash = "sha256:01284878c966508b6d6f1dbff9b6fa607bc062d8261c5c7253cb285b06422a7f", size = 142004, upload-time = "2026-02-09T19:09:40.561Z" },
-]
-
 [[package]]
 name = "grpcio"
 version = "1.78.1"
@@ -1387,15 +1358,6 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/81/db/e655086b7f3a705df045bf0933bdd9c2f79bb3c97bfef1384598bb79a217/keyring-25.7.0-py3-none-any.whl", hash = "sha256:be4a0b195f149690c166e850609a477c532ddbfbaed96a404d4e43f8d5e2689f", size = 39160, upload-time = "2025-11-16T16:26:08.402Z" },
 ]
 
-[[package]]
-name = "logfire-api"
-version = "4.25.0"
-source = { registry = "https://pypi.org/simple" }
-sdist = { url = "https://files.pythonhosted.org/packages/94/5c/026cec30d85394aec8f5f12d70edbe2d706837bc9a411bd71a542cedae50/logfire_api-4.25.0.tar.gz", hash = "sha256:7562d5adfe3987291039dddb21947c86cb9d832d068c87d9aa23db86ef07095b", size = 75853, upload-time = "2026-02-19T15:27:29.518Z" }
-wheels = [
-    { url = "https://files.pythonhosted.org/packages/e2/39/83414c0fadb4f11f90e6b80b631aa79f62a605664f0c4693e2ebc7ee73f3/logfire_api-4.25.0-py3-none-any.whl", hash = "sha256:0d607eb09ef5426e26f376ff277a8d401bc5b7b4178ea66db404e13c368494cf", size = 120473, upload-time = "2026-02-19T15:27:25.832Z" },
-]
-
 [[package]]
 name = "loguru"
 version = "0.7.3"
@@ -2204,24 +2166,6 @@ timezone = [
     { name = "tzdata", marker = "sys_platform == 'win32'" },
 ]
 
-[[package]]
-name = "pydantic-ai-slim"
-version = "1.63.0"
-source = { registry = "https://pypi.org/simple" }
-dependencies = [
-    { name = "genai-prices" },
-    { name = "griffelib" },
-    { name = "httpx" },
-    { name = "opentelemetry-api" },
-    { name = "pydantic" },
-    { name = "pydantic-graph" },
-    { name = "typing-inspection" },
-]
-sdist = { url = "https://files.pythonhosted.org/packages/da/6d/2b5c0c60b42e6af49830f6a09b5d38fecdb1f20d9659152691eba95613b4/pydantic_ai_slim-1.63.0.tar.gz", hash = "sha256:9377afecdfe4bc17f5c9ed72c758e460703ac5876931aa2f18ace8ac0e69312a", size = 426862, upload-time = "2026-02-23T17:56:36.215Z" }
-wheels = [
-    { url = "https://files.pythonhosted.org/packages/f2/ca/c4e39eec1cff5a294b64313a8a959b38d326819e0f0a41f48e61ce019a22/pydantic_ai_slim-1.63.0-py3-none-any.whl", hash = "sha256:ed393b0f871b748171f65bec5191c3025b5abb8a4fc616afee17eb9dc2dfa15d", size = 554190, upload-time = "2026-02-23T17:56:29.533Z" },
-]
-
 [[package]]
 name = "pydantic-core"
 version = "2.41.5"
@@ -2293,23 +2237,6 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/f7/07/34573da085946b6a313d7c42f82f16e8920bfd730665de2d11c0c37a74b5/pydantic_core-2.41.5-graalpy312-graalpy250_312_native-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:76d0819de158cd855d1cbb8fcafdf6f5cf1eb8e470abe056d5d161106e38062b", size = 2139017, upload-time = "2025-11-04T13:42:59.471Z" },
 ]
 
-[[package]]
-name = "pydantic-evals"
-version = "1.63.0"
-source = { registry = "https://pypi.org/simple" }
-dependencies = [
-    { name = "anyio" },
-    { name = "logfire-api" },
-    { name = "pydantic" },
-    { name = "pydantic-ai-slim" },
-    { name = "pyyaml" },
-    { name = "rich" },
-]
-sdist = { url = "https://files.pythonhosted.org/packages/99/43/21b6ddf65b56f7401c344f98e4e6258a02d2868c8a52a8b79c0e0e701029/pydantic_evals-1.63.0.tar.gz", hash = "sha256:eed56a7192e07c8be8cf16e53bb2ef652b4f7f7b8527650ac45fde865a4ecf9d", size = 56365, upload-time = "2026-02-23T17:56:37.71Z" }
-wheels = [
-    { url = "https://files.pythonhosted.org/packages/9b/f2/7174ad6abca2457e35a1b902ca4fa78aa8ee72e4ec2e9cd5dc8904014ec9/pydantic_evals-1.63.0-py3-none-any.whl", hash = "sha256:2e92a3af579a5670b2babf2044081d0ef99ab5a9ef141972616d71fd7e5bfd0e", size = 67279, upload-time = "2026-02-23T17:56:31.008Z" },
-]
-
 [[package]]
 name = "pydantic-extra-types"
 version = "2.11.0"
@@ -2323,21 +2250,6 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/fe/17/fabd56da47096d240dd45ba627bead0333b0cf0ee8ada9bec579287dadf3/pydantic_extra_types-2.11.0-py3-none-any.whl", hash = "sha256:84b864d250a0fc62535b7ec591e36f2c5b4d1325fa0017eb8cda9aeb63b374a6", size = 74296, upload-time = "2025-12-31T16:18:26.38Z" },
 ]
 
-[[package]]
-name = "pydantic-graph"
-version = "1.63.0"
-source = { registry = "https://pypi.org/simple" }
-dependencies = [
-    { name = "httpx" },
-    { name = "logfire-api" },
-    { name = "pydantic" },
-    { name = "typing-inspection" },
-]
-sdist = { url = "https://files.pythonhosted.org/packages/7a/c8/aa3cb56552562b799f31e9de291c8bd88306308cfc9647d220dfff2bea18/pydantic_graph-1.63.0.tar.gz", hash = "sha256:5fd98bb22fa6181f0357a6ffad38a3214af12868bd46492d6456c5db434466b4", size = 58528, upload-time = "2026-02-23T17:56:39.118Z" }
-wheels = [
-    { url = "https://files.pythonhosted.org/packages/a4/1c/8dcae24c824dd2690fbe7375083b369b10ed1ad773e2b9d1122bb6c0fcdc/pydantic_graph-1.63.0-py3-none-any.whl", hash = "sha256:d9b7a387116f358d470c042b07aa08125cadfcfa8c08ef01769746a489aef0d5", size = 72353, upload-time = "2026-02-23T17:56:32.304Z" },
-]
-
 [[package]]
 name = "pydantic-settings"
 version = "2.13.1"
diff --git a/docs/NOTE-FORMAT.md b/docs/NOTE-FORMAT.md
index be1ebb718..2d666f08a 100644
--- a/docs/NOTE-FORMAT.md
+++ b/docs/NOTE-FORMAT.md
@@ -166,6 +166,17 @@ Explicit relations:
 - 'in response to' [[Incident Review]]
 ```
 
+When the text before `[[` is ordinary prose rather than a relation type, add
+`#bm:links_to` to force the reference to use the implicit `links_to` relation:
+
+```markdown
+- Mother [[Alice]] #bm:links_to
+```
+
+The directive is parser metadata and is not included in the observation or
+relation context. It is especially useful when a single-token prose prefix
+would otherwise be interpreted as an explicit relation type.
+
 Bare wiki links and prose list items create implicit `links_to` relations:
 
 ```markdown
diff --git a/integrations/hermes/__init__.py b/integrations/hermes/__init__.py
index 81c821168..8941cab7f 100644
--- a/integrations/hermes/__init__.py
+++ b/integrations/hermes/__init__.py
@@ -39,7 +39,7 @@
 from agent.memory_provider import MemoryProvider
 from tools.registry import tool_error
 
-__version__ = "0.22.1"
+__version__ = "0.23.0"
 
 logger = logging.getLogger("hermes.memory.basic-memory")
 
diff --git a/integrations/hermes/plugin.yaml b/integrations/hermes/plugin.yaml
index 8f6158104..cfce11bf1 100644
--- a/integrations/hermes/plugin.yaml
+++ b/integrations/hermes/plugin.yaml
@@ -1,5 +1,5 @@
 name: basic-memory
-version: 0.22.1
+version: 0.23.0
 description: "Basic Memory — persistent knowledge graph backed by the basic-memory MCP server"
 manifest_version: 2
 api_version: 1
diff --git a/integrations/openclaw/package.json b/integrations/openclaw/package.json
index 5a09c56fe..884a9949e 100644
--- a/integrations/openclaw/package.json
+++ b/integrations/openclaw/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@basicmemory/openclaw-basic-memory",
-  "version": "0.22.1",
+  "version": "0.23.0",
   "type": "module",
   "main": "./dist/index.js",
   "types": "./dist/index.d.ts",
diff --git a/justfile b/justfile
index 13842d03d..da1da1c12 100644
--- a/justfile
+++ b/justfile
@@ -706,6 +706,8 @@ release version:
         .claude-plugin/marketplace.json \
         plugins/claude-code/.claude-plugin/plugin.json \
         plugins/claude-code/.claude-plugin/marketplace.json \
+        plugins/claude-code/hooks/session_start.py \
+        plugins/claude-code/hooks/pre_compact.py \
         plugins/codex/.codex-plugin/plugin.json \
         plugins/codex/hooks/session_start.py \
         plugins/codex/hooks/pre_compact.py \
@@ -826,6 +828,8 @@ beta version:
         .claude-plugin/marketplace.json \
         plugins/claude-code/.claude-plugin/plugin.json \
         plugins/claude-code/.claude-plugin/marketplace.json \
+        plugins/claude-code/hooks/session_start.py \
+        plugins/claude-code/hooks/pre_compact.py \
         plugins/codex/.codex-plugin/plugin.json \
         plugins/codex/hooks/session_start.py \
         plugins/codex/hooks/pre_compact.py \
diff --git a/plugins/claude-code/.claude-plugin/marketplace.json b/plugins/claude-code/.claude-plugin/marketplace.json
index 33ad58788..ec6af13e0 100644
--- a/plugins/claude-code/.claude-plugin/marketplace.json
+++ b/plugins/claude-code/.claude-plugin/marketplace.json
@@ -6,14 +6,14 @@
   },
   "metadata": {
     "description": "Official plugins from Basic Machines for knowledge management and AI-assisted development",
-    "version": "0.22.1"
+    "version": "0.23.0"
   },
   "plugins": [
     {
       "name": "basic-memory",
       "source": "./",
       "description": "The bridge between Claude's working memory and Basic Memory's durable knowledge graph \u2014 session briefings, pre-compaction checkpoints, and capture reflexes",
-      "version": "0.22.1",
+      "version": "0.23.0",
       "author": {
         "name": "Basic Machines"
       },
diff --git a/plugins/claude-code/.claude-plugin/plugin.json b/plugins/claude-code/.claude-plugin/plugin.json
index 35283d3be..9f4f89661 100644
--- a/plugins/claude-code/.claude-plugin/plugin.json
+++ b/plugins/claude-code/.claude-plugin/plugin.json
@@ -1,7 +1,7 @@
 {
   "name": "basic-memory",
   "description": "The bridge between Claude's working memory and Basic Memory's durable knowledge graph \u2014 session briefings, pre-compaction checkpoints, and capture reflexes",
-  "version": "0.22.1",
+  "version": "0.23.0",
   "author": {
     "name": "Basic Machines"
   },
diff --git a/plugins/claude-code/hooks/pre_compact.py b/plugins/claude-code/hooks/pre_compact.py
index bdb002fd4..69c550cea 100755
--- a/plugins/claude-code/hooks/pre_compact.py
+++ b/plugins/claude-code/hooks/pre_compact.py
@@ -1,7 +1,14 @@
 #!/usr/bin/env -S uv run --quiet --script
 # /// script
 # requires-python = ">=3.12"
-# dependencies = ["basic-memory>=0.22.1"]
+# dependencies = [
+#     "basic-memory>=0.23.0",
+#     # Direct pre-release pin so old uv resolves basic-memory (which
+#     # requires this exact beta transitively) without enabling
+#     # pre-releases for basic-memory itself. Keep in lockstep with
+#     # core pyproject's fastmcp pin.
+#     "fastmcp==4.0.0b1",
+# ]
 # ///
 """PreCompact hook — the entire hook. All logic (settings resolution, the
 extractive checkpoint note, lifecycle-envelope capture) lives in the released
diff --git a/plugins/claude-code/hooks/session_start.py b/plugins/claude-code/hooks/session_start.py
index 2dc789223..d3c9203fc 100755
--- a/plugins/claude-code/hooks/session_start.py
+++ b/plugins/claude-code/hooks/session_start.py
@@ -1,7 +1,14 @@
 #!/usr/bin/env -S uv run --quiet --script
 # /// script
 # requires-python = ">=3.12"
-# dependencies = ["basic-memory>=0.22.1"]
+# dependencies = [
+#     "basic-memory>=0.23.0",
+#     # Direct pre-release pin so old uv resolves basic-memory (which
+#     # requires this exact beta transitively) without enabling
+#     # pre-releases for basic-memory itself. Keep in lockstep with
+#     # core pyproject's fastmcp pin.
+#     "fastmcp==4.0.0b1",
+# ]
 # ///
 """SessionStart hook — the entire hook. All logic (settings resolution, the
 context brief, lifecycle-envelope capture) lives in the released basic-memory
diff --git a/plugins/claude-code/hooks/test_claude_pre_compact.py b/plugins/claude-code/hooks/test_claude_pre_compact.py
index 6c44ceec6..ec2cb62c2 100644
--- a/plugins/claude-code/hooks/test_claude_pre_compact.py
+++ b/plugins/claude-code/hooks/test_claude_pre_compact.py
@@ -119,7 +119,7 @@ def test_dependency_floor_matches_package_version() -> None:
     # scripts/update_versions.py bumps this line at release; drift between the
     # script floor and the package version fails here, before a release lands.
     text = SCRIPT.read_text(encoding="utf-8")
-    floors = re.findall(r'^# dependencies = \["basic-memory>=([^"]+)"\]$', text, re.MULTILINE)
+    floors = re.findall(r'^#     "basic-memory>=([^"]+)",$', text, re.MULTILINE)
 
     assert floors == [__version__]
 
diff --git a/plugins/claude-code/hooks/test_claude_session_start.py b/plugins/claude-code/hooks/test_claude_session_start.py
index 8d5c9110e..460c19efc 100644
--- a/plugins/claude-code/hooks/test_claude_session_start.py
+++ b/plugins/claude-code/hooks/test_claude_session_start.py
@@ -119,7 +119,7 @@ def test_dependency_floor_matches_package_version() -> None:
     # scripts/update_versions.py bumps this line at release; drift between the
     # script floor and the package version fails here, before a release lands.
     text = SCRIPT.read_text(encoding="utf-8")
-    floors = re.findall(r'^# dependencies = \["basic-memory>=([^"]+)"\]$', text, re.MULTILINE)
+    floors = re.findall(r'^#     "basic-memory>=([^"]+)",$', text, re.MULTILINE)
 
     assert floors == [__version__]
 
diff --git a/plugins/codex/.codex-plugin/plugin.json b/plugins/codex/.codex-plugin/plugin.json
index 2da0fe369..bec99876c 100644
--- a/plugins/codex/.codex-plugin/plugin.json
+++ b/plugins/codex/.codex-plugin/plugin.json
@@ -1,6 +1,6 @@
 {
   "name": "codex",
-  "version": "0.22.1",
+  "version": "0.23.0",
   "description": "A Codex-native bridge to Basic Memory for durable engineering context, decisions, and resumable checkpoints.",
   "author": {
     "name": "Basic Machines",
diff --git a/plugins/codex/hooks/pre_compact.py b/plugins/codex/hooks/pre_compact.py
index f926d4969..d4fb4de88 100755
--- a/plugins/codex/hooks/pre_compact.py
+++ b/plugins/codex/hooks/pre_compact.py
@@ -2,7 +2,11 @@
 # /// script
 # requires-python = ">=3.12"
 # dependencies = [
-#     "basic-memory @ git+https://github.com/basicmachines-co/basic-memory@2f58b81b5e00cf4cecf12ce595ec6b500d6575a8",
+#     "basic-memory @ git+https://github.com/basicmachines-co/basic-memory@v0.23.0",
+#     # Direct pre-release pin so old uv resolves basic-memory (which
+#     # requires this exact beta transitively) without enabling
+#     # pre-releases broadly. Keep in lockstep with core pyproject.
+#     "fastmcp==4.0.0b1",
 # ]
 # ///
 """PreCompact hook launcher backed by a pinned Basic Memory revision.
diff --git a/plugins/codex/hooks/session_start.py b/plugins/codex/hooks/session_start.py
index e91cf9a6c..a7b0e271a 100755
--- a/plugins/codex/hooks/session_start.py
+++ b/plugins/codex/hooks/session_start.py
@@ -2,7 +2,11 @@
 # /// script
 # requires-python = ">=3.12"
 # dependencies = [
-#     "basic-memory @ git+https://github.com/basicmachines-co/basic-memory@2f58b81b5e00cf4cecf12ce595ec6b500d6575a8",
+#     "basic-memory @ git+https://github.com/basicmachines-co/basic-memory@v0.23.0",
+#     # Direct pre-release pin so old uv resolves basic-memory (which
+#     # requires this exact beta transitively) without enabling
+#     # pre-releases broadly. Keep in lockstep with core pyproject.
+#     "fastmcp==4.0.0b1",
 # ]
 # ///
 """SessionStart hook launcher backed by a pinned Basic Memory revision.
diff --git a/scripts/update_versions.py b/scripts/update_versions.py
index e3e3d4cb0..3cff5a982 100644
--- a/scripts/update_versions.py
+++ b/scripts/update_versions.py
@@ -153,8 +153,8 @@ def _update_packages(version: str, *, dry_run: bool) -> None:
     for script in HOOK_SCRIPTS:
         update_text(
             script,
-            r'^# dependencies = \["basic-memory>=[^"]+"\]$',
-            f'# dependencies = ["basic-memory>={version}"]',
+            r'^#     "basic-memory>=[^"]+",$',
+            f'#     "basic-memory>={version}",',
             dry_run=dry_run,
         )
     update_text(
diff --git a/scripts/validate_claude_plugin.py b/scripts/validate_claude_plugin.py
index e8fe10059..0af19e7ad 100644
--- a/scripts/validate_claude_plugin.py
+++ b/scripts/validate_claude_plugin.py
@@ -137,6 +137,20 @@ def validate_claude_plugin(plugin_dir: Path) -> None:
     for event in REQUIRED_HOOK_EVENTS:
         if event not in hooks:
             raise SystemExit(f"hooks/hooks.json: missing {event} hook")
+    # Trigger: the shims pin core's exact fastmcp beta so older uv can resolve
+    #   basic-memory (pre-release transitives are refused there) without
+    #   enabling pre-releases for basic-memory itself.
+    # Why: nothing else keeps the shim pin and core's pyproject in lockstep,
+    #   and drift makes shim resolution conflict — which the fail-open contract
+    #   turns into a silent hook no-op.
+    # Outcome: validation fails loudly when core moves its fastmcp pin without
+    #   the shims following.
+    core_pyproject = (ROOT / "pyproject.toml").read_text(encoding="utf-8")
+    core_match = re.search(r'^\s*"fastmcp==([^"]+)",$', core_pyproject, re.MULTILINE)
+    if not core_match:
+        raise SystemExit("pyproject.toml: no exact fastmcp pin found")
+    core_fastmcp_pin = core_match.group(1)
+
     for rel in REQUIRED_HOOK_SCRIPTS:
         script = plugin_dir / rel
         if not script.exists():
@@ -145,9 +159,17 @@ def validate_claude_plugin(plugin_dir: Path) -> None:
             raise SystemExit(f"Hook script is not executable: {script}")
         text = script.read_text(encoding="utf-8")
         if "# /// script" not in text or not re.search(
-            r'^# dependencies = \["basic-memory>=[^"]+"\]$', text, re.MULTILINE
+            r'^#     "basic-memory>=[^"]+",$', text, re.MULTILINE
         ):
             raise SystemExit(f"Hook script missing PEP 723 basic-memory floor: {script}")
+        shim_pins = re.findall(r'^#     "fastmcp==([^"]+)",$', text, re.MULTILINE)
+        if len(shim_pins) != 1:
+            raise SystemExit(f"Hook script missing exact fastmcp pin: {script}")
+        if shim_pins[0] != core_fastmcp_pin:
+            raise SystemExit(
+                f"{script}: fastmcp pin {shim_pins[0]} does not match "
+                f"core pyproject pin {core_fastmcp_pin}"
+            )
 
     # --- Output style ---
     output_style = plugin_dir / "output-styles/basic-memory.md"
diff --git a/server.json b/server.json
index ef2ac2ab3..3fc7be291 100644
--- a/server.json
+++ b/server.json
@@ -6,12 +6,12 @@
     "url": "https://github.com/basicmachines-co/basic-memory.git",
     "source": "github"
   },
-  "version": "0.22.1",
+  "version": "0.23.0",
   "packages": [
     {
       "registryType": "pypi",
       "identifier": "basic-memory",
-      "version": "0.22.1",
+      "version": "0.23.0",
       "runtimeHint": "uvx",
       "runtimeArguments": [
         {
diff --git a/src/basic_memory/__init__.py b/src/basic_memory/__init__.py
index 10bd93938..c627c9da0 100644
--- a/src/basic_memory/__init__.py
+++ b/src/basic_memory/__init__.py
@@ -1,7 +1,7 @@
 """basic-memory - Local-first knowledge management combining Zettelkasten with knowledge graphs"""
 
 # Package version - updated by release automation
-__version__ = "0.22.1"
+__version__ = "0.23.0"
 
 # API version for FastAPI - independent of package version
 __api_version__ = "v2"
diff --git a/src/basic_memory/cli/commands/cloud/webdav_transfer.py b/src/basic_memory/cli/commands/cloud/webdav_transfer.py
index 0c824f1c3..26cfd22f7 100644
--- a/src/basic_memory/cli/commands/cloud/webdav_transfer.py
+++ b/src/basic_memory/cli/commands/cloud/webdav_transfer.py
@@ -56,7 +56,7 @@
     upload_file,
 )
 from basic_memory.ignore_utils import load_gitignore_patterns, should_ignore_path
-from basic_memory.mcp.async_client import get_cloud_proxy_client
+from basic_memory.mcp.async_client import get_cloud_control_plane_client
 
 console = Console()
 
@@ -123,7 +123,13 @@ async def webdav_project_diff(
     Raises:
         WebdavError: If the project cannot be listed.
     """
-    cm_factory = client_cm_factory or partial(get_cloud_proxy_client, workspace=workspace_id)
+    # The tenant WebDAV surface is mounted at the cloud app root (/webdav),
+    # not behind /proxy: the proxy catch-alls forward to the per-tenant core
+    # instance, which serves no WebDAV routes, so the proxy client 404s on
+    # every listing (cloud#1816).
+    cm_factory = client_cm_factory or partial(
+        get_cloud_control_plane_client, workspace=workspace_id
+    )
     async with cm_factory() as client:
         remote_files = await list_project_files(client, project)
 
@@ -188,7 +194,11 @@ async def webdav_project_transfer(
             console.print(f"  [dim]{transfer.describe()}[/dim]")
         return
 
-    cm_factory = client_cm_factory or partial(get_cloud_proxy_client, workspace=workspace_id)
+    # Root-mounted /webdav needs the control-plane base — see the note in
+    # webdav_project_diff (cloud#1816).
+    cm_factory = client_cm_factory or partial(
+        get_cloud_control_plane_client, workspace=workspace_id
+    )
     async with cm_factory() as client:
         if direction == "push":
             transfers, appeared = await _drop_appeared_on_cloud(client, project, transfers)
diff --git a/src/basic_memory/markdown/plugins.py b/src/basic_memory/markdown/plugins.py
index 5dfd3536d..1a41ccc63 100644
--- a/src/basic_memory/markdown/plugins.py
+++ b/src/basic_memory/markdown/plugins.py
@@ -13,6 +13,15 @@
 # those bracket prefixes stay ordinary content (issue #1219).
 _TIMESTAMP_VALUE = r"\d{1,3}:\d{2}(?::\d{2})?(?:[.,]\d{1,3})?"
 _TIMESTAMP_CATEGORY = re.compile(rf"^{_TIMESTAMP_VALUE}(?:\s+-\s+{_TIMESTAMP_VALUE})?$")
+_LINKS_TO_DIRECTIVE = re.compile(r"\s+#bm:links_to\s*$")
+
+
+def remove_links_to_directive(content: str) -> tuple[str, bool]:
+    """Remove an exact terminal ``#bm:links_to`` directive from content."""
+    match = _LINKS_TO_DIRECTIVE.search(content)
+    if not match:
+        return content, False
+    return content[: match.start()].rstrip(), True
 
 
 def _is_task_marker_category(category: str) -> bool:
@@ -47,6 +56,7 @@ def is_observation(token: Token) -> bool:
         return False
     # Use token.tag which contains the actual content for test tokens, fallback to content
     content = (token.tag or token.content).strip()
+    content, _ = remove_links_to_directive(content)
     if not content:  # pragma: no cover
         return False
     # if it's a markdown_task, return false
@@ -74,6 +84,7 @@ def parse_observation(token: Token) -> Dict[str, Any]:
 
     # Use token.tag which contains the actual content for test tokens, fallback to content
     content = (token.tag or token.content).strip()
+    content, _ = remove_links_to_directive(content)
 
     # Parse [category] with regex; a timestamp-shaped prefix is not a category, so a
     # hashtag-promoted transcript line keeps its timecode inside the content instead.
@@ -325,17 +336,19 @@ def relation_rule(state: Any) -> None:
 
             # Only process inline tokens
             if token.type == "inline":
+                content = token.tag or token.content
+                content_without_directive, has_directive = remove_links_to_directive(content)
+
                 # Check for explicit relations in list items
-                if in_list_item and is_explicit_relation(token):
+                if in_list_item and not has_directive and is_explicit_relation(token):
                     rel = parse_relation(token)
                     if rel:
                         token.meta["relations"] = [rel]
 
                 # Always check for inline links in any text
                 else:
-                    content = token.tag or token.content
                     if "[[" in content:
-                        rels = parse_inline_relations(content)
+                        rels = parse_inline_relations(content_without_directive)
                         if rels:
                             token.meta["relations"] = token.meta.get("relations", []) + rels
 
diff --git a/src/basic_memory/mcp/tools/write_note.py b/src/basic_memory/mcp/tools/write_note.py
index edd823635..7b521ccc8 100644
--- a/src/basic_memory/mcp/tools/write_note.py
+++ b/src/basic_memory/mcp/tools/write_note.py
@@ -109,12 +109,15 @@ async def write_note(
         - Explicit: `- relation_type [[Entity]] (optional context)`
         - Quoted: `- "multi word relation type" [[Entity]] (optional context)`
         - Quoted: `- 'multi word relation type' [[Entity]] (optional context)`
+        - Disambiguation: Add `#bm:links_to` when prose before `[[Entity]]`
+          must not be treated as a single-token relation type
         - Inline: Any other `[[Entity]]` reference creates a `links_to` relation
 
         Examples:
         `- depends_on [[Content Parser]] (Need for semantic extraction)`
         `- "based on" [[Design Notes]]`
         `- 'in response to' [[Incident Review]]`
+        `- Mother [[Alice]] #bm:links_to`
         `- implements [[Search Spec]] (Initial implementation)`
         `- This feature extends [[Base Design]] and uses [[Core Utils]]`
 
diff --git a/tests/markdown/test_relation_edge_cases.py b/tests/markdown/test_relation_edge_cases.py
index e38112bcc..f3a32ee59 100644
--- a/tests/markdown/test_relation_edge_cases.py
+++ b/tests/markdown/test_relation_edge_cases.py
@@ -2,7 +2,13 @@
 
 from markdown_it import MarkdownIt
 
-from basic_memory.markdown.plugins import relation_plugin, parse_relation, parse_inline_relations
+from basic_memory.markdown.plugins import (
+    observation_plugin,
+    relation_plugin,
+    parse_relation,
+    parse_inline_relations,
+)
+from basic_memory.markdown.entity_parser import parse
 from basic_memory.markdown.schemas import Relation
 
 
@@ -247,3 +253,58 @@ def test_bare_list_wikilink_is_inline_link_not_default_explicit_relation():
 
     assert token.meta["relations"] == [{"type": "links_to", "target": "Target", "context": None}]
     assert parse_relation(token) is None
+
+
+def test_links_to_directive_forces_implicit_relations():
+    """The terminal directive disambiguates a single-token relation prefix."""
+    md = MarkdownIt().use(relation_plugin)
+
+    tokens = md.parse("- Mother [[Alice]] #bm:links_to")
+    token = next(t for t in tokens if t.type == "inline")
+    assert token.meta["relations"] == [
+        {"type": "links_to", "target": "Alice", "context": None}
+    ]
+
+    tokens = md.parse("- Mentions [[Alice]] and [[Bob]] #bm:links_to")
+    token = next(t for t in tokens if t.type == "inline")
+    assert token.meta["relations"] == [
+        {"type": "links_to", "target": "Alice", "context": None},
+        {"type": "links_to", "target": "Bob", "context": None},
+    ]
+
+
+def test_links_to_directive_is_terminal_and_explicit_relations_are_unchanged():
+    """Only the exact terminal directive changes relation interpretation."""
+    md = MarkdownIt().use(relation_plugin)
+
+    tokens = md.parse("- spouse_of [[Alice]]")
+    token = next(t for t in tokens if t.type == "inline")
+    assert token.meta["relations"][0]["type"] == "spouse_of"
+
+    tokens = md.parse("- Mother [[Alice]] #bm:links_to later")
+    token = next(t for t in tokens if t.type == "inline")
+    assert token.meta["relations"][0]["type"] == "links_to"
+
+
+def test_links_to_directive_is_not_an_observation_tag():
+    """The directive is syntax, not a note tag or indexed observation text."""
+    md = MarkdownIt().use(observation_plugin).use(relation_plugin)
+
+    tokens = md.parse("- Mother [[Alice]] #bm:links_to")
+    token = next(t for t in tokens if t.type == "inline")
+    assert "observation" not in token.meta
+
+
+def test_links_to_directive_preserves_source_and_observation_content():
+    """The directive stays in source while remaining outside indexed semantics."""
+    source = "- [note] Mother [[Alice]] #bm:links_to\n"
+    parsed = parse(source)
+
+    assert parsed.content == source
+    assert len(parsed.observations) == 1
+    assert parsed.observations[0].category == "note"
+    assert parsed.observations[0].content == "Mother [[Alice]]"
+    assert parsed.observations[0].tags is None
+    assert len(parsed.relations) == 1
+    assert parsed.relations[0].type == "links_to"
+    assert parsed.relations[0].target == "Alice"
diff --git a/tests/test_claude_plugin_hooks.py b/tests/test_claude_plugin_hooks.py
index 2c3fea411..552bd173c 100644
--- a/tests/test_claude_plugin_hooks.py
+++ b/tests/test_claude_plugin_hooks.py
@@ -80,7 +80,7 @@ def test_claude_script_floor_matches_released_version(event: str, script_name: s
     # Release drift between the PEP 723 floor and the package version fails
     # here (and in each script's co-located test) before a release lands.
     text = (REPO_ROOT / "plugins/claude-code/hooks" / script_name).read_text(encoding="utf-8")
-    floors = re.findall(r'^# dependencies = \["basic-memory>=([^"]+)"\]$', text, re.MULTILINE)
+    floors = re.findall(r'^#     "basic-memory>=([^"]+)",$', text, re.MULTILINE)
 
     assert floors == [CURRENT_VERSION]
 
diff --git a/tests/test_codex_plugin_package.py b/tests/test_codex_plugin_package.py
index 8e2eb3d82..4c9f5b32c 100644
--- a/tests/test_codex_plugin_package.py
+++ b/tests/test_codex_plugin_package.py
@@ -386,3 +386,21 @@ def test_pr_create_skill_delegates_to_current_pr_workflow() -> None:
     assert "never merges" in skill
     assert "Do not enable auto-merge" in skill
     assert "current-head gate" in skill
+
+
+def test_codex_hook_fastmcp_pins_match_core_pyproject() -> None:
+    """The shims pin core's exact fastmcp beta; drift breaks shim resolution.
+
+    Old uv refuses pre-release transitives, so each shim carries the pin as a
+    direct dependency. Nothing rewrites it automatically — this test is the
+    lockstep enforcement when core's pyproject moves its fastmcp pin.
+    """
+    repo_root = Path(__file__).resolve().parents[1]
+    core = re.search(
+        r'^\s*"fastmcp==([^"]+)",$', (repo_root / "pyproject.toml").read_text(), re.MULTILINE
+    )
+    assert core is not None
+    for name in ("session_start.py", "pre_compact.py"):
+        shim = (repo_root / "plugins/codex/hooks" / name).read_text()
+        pins = re.findall(r'^#     "fastmcp==([^"]+)",$', shim, re.MULTILINE)
+        assert pins == [core.group(1)], name
diff --git a/tests/test_read_load_bench.py b/tests/test_read_load_bench.py
index d5aedd2a5..d609edca7 100644
--- a/tests/test_read_load_bench.py
+++ b/tests/test_read_load_bench.py
@@ -10,6 +10,7 @@
 from types import ModuleType
 
 import pytest
+from mcp.types import CallToolResult, TextContent
 
 
 def load_read_load_bench() -> ModuleType:
@@ -26,6 +27,21 @@ def load_read_load_bench() -> ModuleType:
 read_load_bench = load_read_load_bench()
 
 
+def test_result_helpers_use_current_mcp_python_fields() -> None:
+    success = CallToolResult(
+        content=[TextContent(type="text", text="plain response")],
+        structured_content={"result": {"permalink": "notes/example"}},
+    )
+    failure = CallToolResult(
+        content=[TextContent(type="text", text="failed response")],
+        is_error=True,
+    )
+
+    assert read_load_bench.result_payload(success) == {"permalink": "notes/example"}
+    assert read_load_bench.result_text(success) == "plain response"
+    assert read_load_bench.result_text(failure) is None
+
+
 @dataclass
 class RecordingScalarResult:
     version: str
diff --git a/tests/test_update_versions.py b/tests/test_update_versions.py
index d0fbd07c3..06b3cbda9 100644
--- a/tests/test_update_versions.py
+++ b/tests/test_update_versions.py
@@ -35,7 +35,10 @@ def test_parse_version_preserves_python_prerelease_for_non_npm_manifests() -> No
     "#!/usr/bin/env -S uv run --quiet --script\n"
     "# /// script\n"
     '# requires-python = ">=3.12"\n'
-    '# dependencies = ["basic-memory>=0.0.0"]\n'
+    "# dependencies = [\n"
+    '#     "basic-memory>=0.0.0",\n'
+    '#     "fastmcp==4.0.0b1",\n'
+    "# ]\n"
     "# ///\n"
     # Mirrors the real scripts: the docstring mentions a launcher spelling
     # without a version spec, which the anchored updater pattern must skip.