Hypermnesia MCP server
Persistent memory for AI agents — 36 cited neuroscience mechanisms, local-first, hybrid retrieval.
75 stars128 downloads/wk
Reviews
Write oneNobody has reviewed Hypermnesia yet.
If you have run it, two minutes of your experience saves the next person an afternoon.
Hypermnesia tools (63, 6 write)
write = sends, deletes, buys or postsRead from the package source without running it. The installed server may list more.
add_ruleAdd a neuro-symbolic rule to the memory store.
anchorMark a memory as compaction-resistant (heat=1.0).
assess_coverageEvaluate knowledge coverage completeness.
backfill_memoriesAuto-import prior Claude Code conversations into memory.
calibrationScore resolved predictions: Brier score and reliability bands.
change_impactReport memories affected by a commit's code changes.
check_setupVerify the local install: Python, PG driver, DB, extensions, FS.
checkpointSave or restore working state for hippocampal replay.
codebase_analyzeAnalyze codebase and store structure as memories.
consolidatewrite actionRun memory maintenance: decay, compression, CLS, memify.
create_triggerwrite actionCreate a prospective memory trigger.
curate_distillReturn distillation dossiers for the in-session LLM to author lessons from.
curate_wikiReturn structured authoring jobs for the in-session LLM to author.
detect_domainLightweight domain classification.
detect_gapsIdentify knowledge gaps in the memory store.
drill_downNavigate into a fractal memory cluster.
echoexplore_featuresExplore interpretability features.
fixtureforgetwrite actionDelete or soft-delete a memory by ID.
get_causal_chainTrace entity relationships through the knowledge graph.
get_grooming_healthBacklog + staleness for wiki/distillation/promotion grooming.
get_project_storyGenerate a period-based autobiographical narrative.
get_rulesList active neuro-symbolic rules.
get_telemetryReturn per-op counters + read/write ratio (Popper C6).
import_sessionsImport conversation history into the memory store.
ingest_codebaseIngest upstream codebase analysis into Cortex.
ingest_documentIngest a .docx or Confluence export into Cortex.
ingest_findingsIngest an AP findings run (runs/<run_id>/) into Cortex.
ingest_prdIngest a PRD document into Cortex.
lesson_promotionPropose promotion jobs for validated lessons — never promotes itself.
list_domainsOverview of all detected cognitive domains.
memory_statsMemory system diagnostics.
narrativeGenerate project narrative from stored memories.
navigate_memoryNavigate memory space using Successor Representation.
predictRecord a falsifiable prediction before its outcome is known.
query_methodologyReturns cognitive profile for the current domain.
rate_memoryRate a memory as useful or not to update metamemory confidence.
rebuild_profilesFull rescan of all session data to rebuild methodology profiles.
recallRetrieve memories using multi-signal fusion.
recall_hierarchicalRetrieve memories using fractal hierarchy.
recall_skillsRetrieve procedural skills for the current situation.
record_session_endIncremental profile update after a session ends.
rememberStore a memory through the predictive coding write gate.
resolve_predictionSettle an open prediction against the evidence that decided it.
seed_projectBootstrap memory from an existing codebase.
slow_toolsync_instructionsPush top memory insights into CLAUDE.md.
unified_searchFuse Cortex recall with AP code search using reciprocal rank fusion.
validate_memoryValidate memories against current filesystem state.
whyResolve ⟦rcpt:id⟧ injection receipts into presence-in-context evidence.
wiki_adrwrite actionCreate a numbered ADR with auto-incremented sequence.
wiki_get_draftFetch a pending draft with its source claims and kind contract.
wiki_linkAdd a bidirectional link between two wiki pages (Related section).
wiki_listList authored wiki pages, optionally filtered by kind.
wiki_migrateSync the filesystem wiki into PG and reconcile FS-deleted ghosts.
wiki_purgeRe-evaluate and purge wiki pages that fail the current classifier.
wiki_readRead a wiki page by relative path.
wiki_refine_draftwrite actionSubmit refined prose for a draft and record an audit memo.
wiki_reindexRegenerate the wiki table of contents at .generated/INDEX.md.
wiki_renameMove a page and leave a redirect stub at the old path.
wiki_verifyVerify wiki-page symbol citations against AP's code graph.
wiki_writewrite actionAuthor a wiki page using create, append, or replace mode.
Public scan report
scanner v0.1.10 · 2026-10-15 · same rubric, same numbers if you re-run it
- Code scan1291 source files scanned25/25
- –Live reliabilityno gateway calls yet and no remote to proben/a
- –Tool poisoningtools not inspected (local package is not executed); not countedn/a
- Auth qualitylocal package, no credentials required12/15
- Maintenancelast push 0 days ago15/15
- Maintainer identityregistry namespace matches repository owner; GitHub account older than a year8/10
How Hypermnesia compares
Against every graded MCP server, from our own scans and uptime checks. Updated after each scan run.
- Safety score
- 92/100
- higher than 95% of 48,255 servers
- Tools
- 63
- the typical server has 7
- Tools that change things
- 10%
- typical: 0%. Fewer write tools means less an agent can break
- GitHub stars
- 75
- more than 95% of 28,688 servers
- Downloads a week
- 128
- more than 68% of 12,191 servers with a package
What its tools let an agent do: 54 read · 3 create · 6 change
Release history: 6 versions in the registry since 2026-09-15, latest 2026-09-23, typically one every 1 day.
Install Hypermnesia in Claude Code, Cursor or VS Code
claude mcp add hypermnesia-mcp -- uvx hypermnesia-mcp
What the publisher says
From the Hypermnesia repository's README, as published. We do not edit it. Read it on GitHub
<!-- mcp-name: io.github.cdeust/hypermnesia-mcp -->
Memory for AI coding agents that you can hold accountable. Keep decisions, fixes and project context between sessions, and inspect what was retrieved. Runs locally by default. No account, no API key, no server to manage.
Sovereign is what it is today. Everything runs on your machine: a local SQLite file by default, or PostgreSQL + pgvector if you prefer. No LLM in the retrieval loop, and nothing leaves localhost unless you configure an integration that does. Your project's memory is a file you own and can delete.
Cross-platform is how it is built. One stdio MCP server and the same 57 tools on Claude Code, in the Claude Desktop bundle, under Claude Cowork, and on every local stdio MCP host listed in the table below. What differs per host is stated there, not discovered after install.
Eco-responsible is what we are aiming at. Work that never reaches a datacenter is work nobody has to power, and an agent that finds the right context first time re-reads fewer files. We hold that intent to the Green Software Foundation's SCI method, and we publish no CO₂ or energy figure, because we have not measured one. What we do and do not claim ↓
36 neuroscience mechanisms · 57 memory tools · 9 lifecycle hooks · a self-curating per-project wiki — all local, all open-source, MIT.
Install
Claude Code — add the marketplace and install the plugin:
claude plugin marketplace add cdeust/Cortex
claude plugin install hypermnesia-mcpClaude Desktop — download hypermnesia-mcp.mcpb from Releases and open it, or use Settings → Extensions. The bundle carries the tools but no hooks; the MCPB format has none.
Claude Cowork is detected automatically (CLAUDE_ENVIRONMENT=cowork) and uses the local SQLite store. No PostgreSQL required.
Any other stdio MCP host (Codex, Gemini CLI, Cursor, Windsurf, VS Code) launches the same server and gets the same tools. The per-host matrix and launch commands are in Every other MCP host below. Codex has a native package: docs/codex-plugin.md. WSL, TLS client certificates and corporate proxies are covered in docs/deployment-scenarios.md.
The first use creates a local SQLite store under ~/.claude/methodology/. Models are downloaded once when needed and then run offline. The embedding and reranking model files are both fetched on first use. Optional integrations, remote PostgreSQL, and OTLP telemetry use the network only when explicitly configured. PRIVACY.md lists the exact scope.
Shortened. The full README is on GitHub.
Nothing above is checked by us. What we check is on the safety report.
Grade history
- 2026-09-19restoreC → Ascore 92
- 2026-09-19downgradeB → Cscore 66: Whole environment serialized (possible credential exfil); Network call to a raw IP address
Hypermnesia: common questions
- Is Hypermnesia MCP server safe?
- Yes, by our scan: it is graded A (92/100). Read the Hypermnesia safety report
- How do I install Hypermnesia?
- It runs on your machine. Copy the Claude Code, Cursor, VS Code or Claude Desktop config from the install section.
- Does Hypermnesia need an API key?
- Not as far as the registry entry and our scan can tell: no credentials are declared or required.
- Is Hypermnesia maintained?
- The last commit was 20 days ago (2026-10-15). The latest release is v4.23.5.
- What can I use instead of Hypermnesia?
- Servers from other publishers that do the same job: Memorix MCP server, Nautilus Compass Memory MCP server and State Memory MCP server. Compare all Hypermnesia alternatives.
- How does Hypermnesia compare with other MCP servers?
- safety score 92/100 (higher than 95% of 48,255 servers); tools 63 (the typical server has 7); tools that change things 10% (typical: 0%. Fewer write tools means less an agent can break). Full comparison.
Alternatives to Hypermnesia
Same job from other publishers: the closest match first, then the best rated.
MemorixLocal-first project memory with legacy MCP and 2026 discovery compatibility.not reviewedEstablishedB- Nautilus Compass MemoryLocal-first long-term memory for AI agents, with byte-recomputable signed verification receipts.not reviewedEstablishedB
State Memory MCPPersistent workflow graph server tracking tasks, decisions, spatial entities, and blockers.not reviewedEstablishedA
KnowlPersistent memory for AI agents, for code, research and everyday life. Stale facts retire.not reviewedEstablishedA