This release adds 4 notable features for engineering teams evaluating rollout.
✓ No known CVEs patched in this version
Topics
Summary
AI summaryFalkorDB graph data now survives container recreates, provider status probes actual content, and Windows long‑path issues are fixed.
Changes in this release
| Type | Severity | Summary | CVE |
|---|---|---|---|
| Feature | Low |
Additive `ProviderSummary.indexing` list indicates a provider is healthy but mid‑scan Additive `ProviderSummary.indexing` list indicates a provider is healthy but mid‑scan Source: llm_adapter@2026-06-09 Confidence: high |
— |
| Bugfix | Medium |
FalkorDB graph data now survives container recreates via configurable `dataDestination` mount fix FalkorDB graph data now survives container recreates via configurable `dataDestination` mount fix Source: llm_adapter@2026-06-09 Confidence: high |
— |
| Bugfix | Medium |
Empty‑graph targets degrade the global context packet (`ok: false`) with per‑repo restart recovery actions Empty‑graph targets degrade the global context packet (`ok: false`) with per‑repo restart recovery actions Source: llm_adapter@2026-06-09 Confidence: high |
— |
| Bugfix | Medium |
Watcher self‑heal entrypoint guard clears poisoned empty graphs at startup (runner image `0.4.10-ar1`) Watcher self‑heal entrypoint guard clears poisoned empty graphs at startup (runner image `0.4.10-ar1`) Source: llm_adapter@2026-06-09 Confidence: high |
— |
| Bugfix | Medium |
Backend healthcheck now requires a genuine `PONG`; dataset‑`LOADING` replies no longer count as healthy Backend healthcheck now requires a genuine `PONG`; dataset‑`LOADING` replies no longer count as healthy Source: llm_adapter@2026-06-09 Confidence: high |
— |
| Bugfix | Medium |
`sync-skills.py` and `sync-runtime.py` use staged copy‑then‑swap with `\?\` extended‑length path support for crash safety `sync-skills.py` and `sync-runtime.py` use staged copy‑then‑swap with `\?\` extended‑length path support for crash safety Source: llm_adapter@2026-06-09 Confidence: high |
— |
| Bugfix | Medium |
`cgc_status` derives `indexingState` from a read‑only `GRAPH.RO_QUERY` content probe plus watcher scan markers `cgc_status` derives `indexingState` from a read‑only `GRAPH.RO_QUERY` content probe plus watcher scan markers Source: llm_adapter@2026-06-09 Confidence: low |
— |
| Bugfix | Medium |
All watcher compose runs pass `--remove-orphans`, cleaning up watchers of de‑configured repos All watcher compose runs pass `--remove-orphans`, cleaning up watchers of de‑configured repos Source: llm_adapter@2026-06-09 Confidence: low |
— |
| Bugfix | Medium |
`worktree_start` preflight blocks and remedies when Windows hosts lack `LongPathsEnabled` for deep tracked files `worktree_start` preflight blocks and remedies when Windows hosts lack `LongPathsEnabled` for deep tracked files Source: llm_adapter@2026-06-09 Confidence: low |
— |
| Bugfix | Medium |
Provider status probes actual graph content, reporting states like `indexed`, `indexing`, `empty`, or `backend-unreachable` Provider status probes actual graph content, reporting states like `indexed`, `indexing`, `empty`, or `backend-unreachable` Source: llm_adapter@2026-06-09 Confidence: low |
— |
Full changelog
Agents Remember 2.5.0
The CodeGraphContext provider becomes durable and honest about readiness, and the toolchain stops fearing Windows long paths. Born from a real incident: a context packet reported green over a 0-node code graph for three days.
Highlights
- FalkorDB graph data now survives container recreates (configurable
dataDestinationmount fix) - Provider status probes actual graph content:
indexed/indexing/empty/backend-unreachable - Empty-graph targets degrade the global context packet (
ok: false) with per-repo restart recovery actions - New additive
ProviderSummary.indexinglist tells agents a provider is healthy but mid-scan - Watcher self-heal entrypoint guard clears poisoned empty graphs at startup (runner image
0.4.10-ar1) - Crash-safe, long-path-proof skill/runtime sync and a
worktree_startlong-path preflight for native Windows
CGC Provider Durability And Readiness
- FalkorDB data volume binds the backend
dataDestination(default/var/lib/falkordb/data, where FalkorDB v4 actually writes) instead of/data; mount verification follows the setting and mismatched containers are recreated on the next backend start. - The backend healthcheck requires a genuine
PONG, so dataset-LOADINGreplies no longer count as healthy. cgc_statusderivesindexingStatefrom a read-onlyGRAPH.RO_QUERYcontent probe plus watcher scan markers;indexingstays green at every level while busy targets surface in the compact providers summary and the l-01 trust checkpoint doctrine.- All watcher compose runs pass
--remove-orphans, cleaning up watchers of de-configured repos.
Windows Long-Path Hardening
sync-skills.py/sync-runtime.pyuse a staged copy-then-swap with\\?\extended-length path support - a mid-sync crash can no longer gutpackage_data.worktree_startblocks up front with projected path lengths and remedies when a Windows host withoutLongPathsEnabledcannot represent the worktree's deepest tracked files.- The install guide documents the one-time Windows long-path setup; WSL is unaffected.
All changes are backward-compatible. Fixes #47.
Weekly OSS security release digest.
The CVE patches and breaking changes that affected production tools this week. One email, every Sunday.
No spam, unsubscribe anytime.
Share this release
About Agents Remember
All releases →Related context
Related tools
Earlier breaking changes
- vmcp-v2.9.0 Removed tools direct_closeout_preview and direct_closeout_apply.
- vmcp-v2.0.0 `WorkflowKind` enum drops `heavy` and `heavy-task` values.
- vmcp-v2.0.0 `skills_install` no longer accepts a `layout` input; installer performs flat copy.
- vmcp-v2.0.0 Removed workflows W-01-heavy-task-workflow and W-03-chat-task-workflow.
Beta — feedback welcome: [email protected]