Query Markup Documents - On-device hybrid search for markdown files with BM25, vector search, and LLM reranking
Nothing to run from here
Memforge is installed from its publisher's own source and answers where it runs, so this marketplace is not in the path of a single call. There is no address here to send one to, and a panel that pretended otherwise would be showing you an answer we made up. Install it and call it from your own client — the Installation tab has the entry for each one.
100/100
trust score
2
downloads a day
local
runs on your machine
123
downloads in 30d
What it does
Neuroscience-inspired agent memory — sleep cycles, memory revision, knowledge graph, procedural learning, vector search, and MCP integration
Quickstart
# 1 — add it to Claude Code
claude mcp add --scope user --env 'MEMFORGE_URL=${MEMFORGE_URL}' --env 'MEMFORGE_TOKEN=${MEMFORGE_TOKEN}' --transport stdio memforge-mcp '--' npx -y --package=@salishforge/memforge memforge-mcp
# 2 — ask your agent for something these tools do
# memforge_add, memforge_query, memforge_consolidate and 42 more
It needs MEMFORGE_URL, MEMFORGE_TOKEN: this reads them from your environment, so export them first.
Memforge costs nothing on mcprush: there is no plan to choose here, no cap to set and nothing mcprush can bill you for.
Publishernot claimed
SAAll 1 server
salishforge1 server
1 category
Collected from a public index. Nobody has claimed this account, so nothing here was written by its author — claim it if it is yours.
Installation
Where are you running it?
Every route below installs the same thing and ends at the same approval screen. This one runs on your machine: your client starts Memforge as a process under your own user, with your files and your network, so the tool surface below is what it can do to your computer rather than to a server somewhere else. Its scan result is on the Security tab — read the surface before you approve it.
Claude Codedeclared by this listing
This is a public server: you run it yourself and this marketplace is not in the path. You need Node.js — the current LTS (24) is the safe choice; npx comes with it. Claude Code registers it in one command; --scope user makes it available in every project. Claude Code reads MEMFORGE_URL (URL of the MemForge service (e.g. http://localhost:3333)), MEMFORGE_TOKEN (Bearer token for MemForge API authentication) from your environment each time it starts the server, so the value stays out of its config: export them in your shell profile (macOS, Linux), or run setx MEMFORGE_URL …, setx MEMFORGE_TOKEN … and open a new terminal (Windows). The '--' is in quotes so that PowerShell passes it on.
claude mcp add --scope user --env 'MEMFORGE_URL=${MEMFORGE_URL}' --env 'MEMFORGE_TOKEN=${MEMFORGE_TOKEN}' --transport stdio memforge-mcp '--' npx -y --package=@salishforge/memforge memforge-mcpReconnect, or start a new session, and the tools appear in the model’s tool list.
Free on mcprush, and it runs from its publisher’s own package: this marketplace is not in the path, so there is no key of ours to hold and no card to add.
What it asks of you
It reaches a system of yours, so it needs your own credential rather than the publisher’s: MEMFORGE_URL — URL of the MemForge service (e.g. http://localhost:3333); MEMFORGE_TOKEN — Bearer token for MemForge API authentication. You set them in your own client’s config, on your machine — this marketplace never holds it.
What it writes
One config entry your client uses to start the process locally. A local server runs with your file system and your network, which is why it is priced without metering.
What you approve
45 tools, with what each one reads, writes and reaches shown before you agree — the same list on every route above. Read the tool surface.
Tool surface
What the model actually sees. Descriptions are diffed on every release — see version history.
Store a memory event for an agent. Returns the created memory ID.
No parameter schema on file.
Search an agent
No parameter schema on file.
Consolidate hot-tier events into searchable warm-tier memories.
No parameter schema on file.
Proactively surface relevant memories and procedures for a given context.
No parameter schema on file.
Get session resumption context: recent memories, procedures, contradictions.
No parameter schema on file.
Record whether retrieved memories were helpful (positive/negative/neutral).
No parameter schema on file.
Run a sleep cycle: score, triage, revise, maintain graph, reflect.
No parameter schema on file.
Get memory health metrics: importance, confidence, revision velocity.
No parameter schema on file.
Get memory tier statistics: hot, warm, cold counts.
No parameter schema on file.
Retrieve memories in chronological order within an optional time range.
No parameter schema on file.
Search knowledge graph entities (people, systems, organizations, concepts).
No parameter schema on file.
Traverse the knowledge graph from a specific entity. Returns connected nodes and edges.
No parameter schema on file.
Trigger a reflection — synthesizes insights and contradictions from recent memories.
No parameter schema on file.
Retrieve stored reflections (synthesized insights from past reviews).
No parameter schema on file.
Retrieve learned procedures — condition→action rules extracted from reflections.
No parameter schema on file.
Trigger meta-reflection — synthesizes higher-order principles from accumulated first-order reflections.
No parameter schema on file.
Detect and merge duplicate entities in the knowledge graph using trigram similarity.
No parameter schema on file.
Search archived (cold tier) memories. Use for audit, recovery, and compliance. Returns rows still within retention window.
No parameter schema on file.
Restore a cold tier row to warm tier for reactivation. Non-destructive — the cold row is preserved for audit.
No parameter schema on file.
Get an adaptive sleep-cycle recommendation. Advisory only — MemForge has no built-in scheduler. Callers (cron jobs, control planes) read the urgency and reason and decide whether to call memforge_sleep.
No parameter schema on file.
Publish an agent\
No parameter schema on file.
List active procedures shared in a pool, ranked by confidence and corroboration. Use to discover what condition→action rules other agents have learned.
No parameter schema on file.
Discover which pool members know the most about a topic. Returns agents ranked by relevance score with sample matching memories. Use to route questions to the right agent.
No parameter schema on file.
Declare an expertise domain for an agent. Roles are used by expertise discovery and for routing queries in multi-agent systems. Upserts on (agent_id, domain).
No parameter schema on file.
Get all declared expertise roles for an agent, ordered by confidence. Includes both manually declared and auto-detected roles.
No parameter schema on file.
Auto-detect expertise roles from an agent\
No parameter schema on file.
Set or clear the validity window on a warm-tier memory. Memories past valid_until are penalized during sleep cycles.
No parameter schema on file.
Record the outcome of executing a procedure. Success/failure counts drive confidence evolution during sleep cycles.
No parameter schema on file.
Fetch drift-detection report based on recent drift_signals snapshots. Trend classification: stable | degrading | recovering | insufficient_data.
No parameter schema on file.
Mark a namespace as deprecated. Sleep cycles will actively decay importance and confidence of memories in this namespace each cycle, eventually evicting them. Reversible.
No parameter schema on file.
Reverse a namespace deprecation. Future sleep cycles stop decaying its rows.
No parameter schema on file.
List the agent
No parameter schema on file.
Enqueue a dream run — async sleep-cycle job mirroring Anthropic Claude Dreaming. Returns a run id (status=
No parameter schema on file.
Fetch a dream run by id. Useful for polling pending or running cycles.
No parameter schema on file.
List dream runs for an agent. Filter by status or source.
No parameter schema on file.
Request cancellation of a dream run. Pending runs go straight to
No parameter schema on file.
Bridge: export warm-tier rows for an agent/namespace to an Anthropic Memory Store. Requires DREAMS_PROVIDER=anthropic + ANTHROPIC_API_KEY on the server.
No parameter schema on file.
Bridge: import records from an Anthropic Memory Store into warm_tier. Strategies: anthropic-wins (default — overwrite), memforge-wins (insert net-new only), merge (overwrite content, preserve local metadata).
No parameter schema on file.
Bridge: report current sync state for an agent/namespace — known store links, last push/pull timestamps, and a drift indicator (warm rows newer than the most recent push).
No parameter schema on file.
Return the count of warm-tier memories per epistemic_status for an agent. Useful for gauging how much of the knowledge base is well-corroborated (established) vs. newly accepted (provisional) vs. uncertain (contested/inferred/deprecated).
No parameter schema on file.
Explain a warm-tier memory
No parameter schema on file.
Traverse causal relationships from a warm-tier memory. Returns a chain of cause/effect memories with edge strength and confidence.
No parameter schema on file.
Predict probable next events for a context, based on causal edges mined from memory sequences. Probability is a relative ranking signal (confidence-scaled, monotonic in edge strength), not a calibrated probability.
No parameter schema on file.
Retrieve an agent
No parameter schema on file.
Return an agent
No parameter schema on file.
Free
$0 forever
- Every tool, no call limit
- No card, and mcprush charges nothing for it
- Source published under MIT
- Runs on your machine — nothing of it reaches our gateway
- Nothing to cap, because mcprush bills nothing
What counts against your monthly calls
| Tool | Unit | Calls used | Out of the allowance |
|---|
Nothing here is billable: mcprush charges nothing to install Memforge or to call it, at any volume.
Call allowance · this installenforced at the gateway
Monthly allowance, in calls
—
—
Alert at 80% · then the tool returns a normal MCP error
Used this month: —
ATrust score 100 / 100
Release 3.0.0-beta.4 · scanned 14 Oct 2026
Two independent axes, because powerful and malicious are different questions. The grade is threat only. The capability level is blast radius, and it is never a penalty on the grade — it is priced as one subtract-only term in the score, where you can see it.
Threat score100
Client score100
Findings1 threat · 0 capability
Capabilityminimal
Coveragesource
Verificationsource
Surface digest7b0a892ca24870870b265ae2e0805b6a2c83e490c6ca6c90f63d5f672b8cf48d
Score vectorthreat 100 → client 100
| Term | Level | What it prices | Points |
|---|---|---|---|
| capability-exposure | minimal | capability blast radius (minimal) — client exposure if the model is manipulated | −0 |
| verification-discount | source | publisher verification (provenance) — cryptographic build provenance ties the artifact to its source | −0 |
| coverage-honesty | source | inspection depth (source) — how much of the target the scan could see | −0 |
score = clamp(round(threatScore − E_cap − E_ver − E_cov), 0, 100). No model, no opinion, no network call at scoring time: the same methodology version against the same bytes always produces this number, and every line above is a field of the report rather than a summary of it.
What the scan could actually read
A grade is only as meaningful as its coverage, so the scanner publishes its own depth before it publishes its result.
Tool surfaceread
Implementation sourceread
Package metadataread
Live transportnot available
Tools were statically extracted from the published source (19 recovered), not enumerated from a running server. Tool-poisoning, Unicode-smuggling, capability and toxic-flow analysis ran on this inferred surface, but a mis-parsed registration could be missed or mis-attributed, so tool-derived findings are capped below “confirmed”. To grade the real runtime surface, scan the running server: --command "npx -y <package>".
Capability — what it could do if the model were manipulated
Tags derived from each tool’s schema and the implementation, not from what the tool calls itself. minimal is the level these add up to.
| Tool | Capability tags | Why the tag was assigned |
|---|---|---|
| memforge_add | no tags | |
| memforge_query | no tags | |
| memforge_timeline | no tags | |
| memforge_entities | no tags | |
| memforge_graph | no tags | |
| memforge_reflect | no tags | |
| memforge_reflections | no tags | |
| memforge_consolidate | no tags | |
| memforge_procedures | no tags | |
| memforge_sleep | no tags | |
| memforge_health | no tags | |
| memforge_stats | no tags | |
| memforge_feedback | no tags | |
| memforge_meta_reflect | no tags | |
| memforge_dedup_entities | no tags | |
| memforge_active_recall | no tags | |
| memforge_cold_search | no tags | |
| memforge_cold_restore | no tags | |
| memforge_sleep_advisory | no tags |
Toxic-flow graph
The lethal trifecta, checked as a graph rather than as a checklist: untrusted input, a sensitive source and an external sink have to meet before there is a path worth worrying about.
The public result for this release does not print the flow graph, so there is nothing to show here. That is not the same as "no paths were found": what the scan did read is above, under coverage.
Supply chain and provenance
Integrityfirst-seen
Surface digest7b0a892ca2487087
Scanned2026-10-14
This is the first scan of this surface here, so there is nothing yet to compare it against.
Every result on this tab comes from one deterministic pass over the published package — offline, rule by rule, and auditable line by line above. Same methodology version, same bytes, same score.
Release history
Always latest — the install command names no version, so it installs the newest release when it runs; the current release on file is 3.0.0-beta.4. Pin a release below and the command asks for exactly that one.
3.0.0-beta.4
undated
latestNo release note was published with this version.
—
no reviews yet
Who can review
One review per account, from a signed-in account that does not publish this listing. Nothing else is asked — you can say what you think before you install it. Publishers can reply once per review, and a review can be edited or deleted by whoever wrote it.
Sign in to write one
Any signed-in account that does not publish Memforge can review it, once — before installing it or after. A review from an account that has installed it is marked as one.
No reviews yet
Nobody has reviewed this listing. The rating on the card is the mean of the reviews written here and nothing else, so there is no rating until somebody writes the first — which takes a signed-in account that does not publish it, and nothing else.