From bde852f7ef18e81b88fc581099700bc455f3b9ed Mon Sep 17 00:00:00 2001 From: Claude Date: Fri, 12 Jun 2026 05:51:07 +0000 Subject: [PATCH 1/2] docs: mark v0.25.0 as the latest release on the landing page v0.25.0 shipped, so the hero badge, JSON-LD softwareVersion, and the timeline pills move forward: v0.25.0 takes the "Latest release" pill and v0.24.0 joins the plain history trail. https://claude.ai/code/session_01FkHXHcjpWZL2EWn4HGi547 --- docs/index.html | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/docs/index.html b/docs/index.html index ced3fc1..6185864 100644 --- a/docs/index.html +++ b/docs/index.html @@ -26,7 +26,7 @@ "description": "Persistent memory and semantic code intelligence for AI coding agents. Local-first code indexing with 40–70× per-query token reduction, a brain-like synapse layer that learns associations from how you use the codebase, an Obsidian-style graph view, an MCP server, and a built-in install doctor. Works with Claude Code, Cursor, Cline, Continue, and any MCP-compatible agent.", "url": "https://dfrostar.github.io/neuralmind/", "downloadUrl": "https://pypi.org/project/neuralmind/", - "softwareVersion": "0.24.0", + "softwareVersion": "0.25.0", "license": "https://opensource.org/licenses/MIT", "programmingLanguage": "Python", "operatingSystemRequirements": "Python 3.10+", @@ -355,7 +355,7 @@
-

v0.24.0 — latest release · release notes

+

v0.25.0 — latest release · release notes

Persistent memory for AI coding agents

Your agent learns your codebase the way a senior engineer would — what files go together, what you usually edit next, what patterns matter. The memory persists across sessions and surfaces automatically.

100% local. Your code never leaves your machine. Side effect: 40–70× cheaper code questions, measured in CI on every commit.

@@ -584,11 +584,11 @@

What's new

Shipped in small, verifiable increments — every release gated by the CI benchmark.

-
v0.25.0In development
+
v0.25.0Latest release

One learning system: the synapse layer. The old learned_patterns cooccurrence reranker is removed and neuralmind learn becomes an exit-0 deprecation no-op. The Hebbian synapse layer — which already learns continuously from queries, edits, and tool calls, and lets unused edges decay — is now the single learning signal. A 2×2 A/B on the benchmark fixture showed the reranker moved top-k hit rate by 0.0 points whether synapses were on or off (71.7% → 71.7% cold, 83.3% → 83.3% warm), while the synapse layer alone adds +11.6 points. Warm-path behavior is unchanged; the only visible difference is that L3 output no longer prints reranker (+X.XX boost) labels (synapse labels stay). Release notes →

-
v0.24.0Latest release
+
v0.24.0

Memory namespaces & branch isolation. The synapse layer becomes namespace-aware: branch:<name> / personal / shared / ephemeral memory live separately in the same store, so a feature-branch spike can't pollute what the agent learned about main. Recall reads a transparent merged view (active branch 1.0× > personal 0.8× > shared team baseline 0.5×, attributed per-namespace in query --trace), and the new neuralmind memory {inspect,reset,export,import} moves memory as versioned JSON bundles — the team-memory on-ramp. Existing learned memory migrates losslessly into personal. Release notes →

From 4f4073bedd8336822622f96ee3f75ad7451b681f Mon Sep 17 00:00:00 2001 From: Claude Date: Fri, 12 Jun 2026 05:58:05 +0000 Subject: [PATCH 2/2] test: tolerate lagging chroma handle release in the integration fixture tests/test_integration_retrieval.py creates its own TemporaryDirectory hosting a ChromaDB store. Fixture teardown is LIFO, so the directory is removed before conftest's autouse chroma-release fixture runs; the embedder close in initialized_mind usually suffices, but Windows can hold the sqlite handle a beat longer (AV scans, deferred closes), which intermittently failed teardown with WinError 32. Mirror the conftest fixtures: ignore_cleanup_errors=True. https://claude.ai/code/session_01FkHXHcjpWZL2EWn4HGi547 --- tests/test_integration_retrieval.py | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) diff --git a/tests/test_integration_retrieval.py b/tests/test_integration_retrieval.py index 0f85a87..94fad16 100644 --- a/tests/test_integration_retrieval.py +++ b/tests/test_integration_retrieval.py @@ -17,8 +17,15 @@ @pytest.fixture def minimal_project(): - """Create a minimal project with sample graph for testing.""" - with tempfile.TemporaryDirectory() as tmpdir: + """Create a minimal project with sample graph for testing. + + ``ignore_cleanup_errors``: this directory hosts a ChromaDB store, and + fixture teardown is LIFO — the temp dir exits before conftest's + autouse chroma-release fixture runs, so on Windows a still-open (or + just-released, AV-scanned) sqlite handle would otherwise fail the + teardown even though ``initialized_mind`` closes the embedder. + """ + with tempfile.TemporaryDirectory(ignore_cleanup_errors=True) as tmpdir: project_path = Path(tmpdir) graphify_out = project_path / "graphify-out" graphify_out.mkdir()