Marrow
MCPB · MARROW-MCP.MCPB
Shared memory for parallel AI coding agents. Local, free, with rooms and file claims.
Available components
Last scanned (11 hours ago)
Scores higher than 94% of MCPB servers
Category Memory and Knowledge
How this component scores in each security and reliability category. Every signal is checked automatically from public evidence about the published package, including repeated runs of it in an isolated sandbox, and we only credit what we can confirm. How we score → Why this is hard to score →
Supply Chain Security0
- No malware scan is available for this kind of package: the supply-chain vendors we use do not cover it. This is a permanent gap in our coverage, not a finding about the package.Unverified
- Known CVEs could not be checked: we found no dependency list to check in this bundle: either it runs a compiled binary, whose bundled manifests we do not read, or none of the package.json, requirements.txt, pyproject.toml or uv.lock files it ships for its runtime declares a dependency we can read (a requirements.txt that only points at a setup.py project, a local path, or a nested requirements file we cannot follow counts as none).Unverified
- Install-script risk not yet assessed.Unverified
- Dependency health could not be checked: we found no dependency list to check in this bundle: either it runs a compiled binary, whose bundled manifests we do not read, or none of the package.json, requirements.txt, pyproject.toml or uv.lock files it ships for its runtime declares a dependency we can read (a requirements.txt that only points at a setup.py project, a local path, or a nested requirements file we cannot follow counts as none).Unverified
Provenance & Transparency100
- Source repository is publicly reachable at the declared URL. View diagnostics → Pass
- Cryptographically verified build provenance (signed, bound to aryawidjaja/marrow). View diagnostics → Pass
- Clear OSI-approved license (AGPL-3.0-only).Pass
- Actively maintained (last published 69 days ago).Pass
- Publishes a security disclosure policy (SECURITY.md).Pass
Schema Quality & AI Usability78
- 100% of prompts and resources have a non-trivial description (not blank, and not just the item's name).Pass
- AI-judged instruction clarity (good).Pass
- Context-footprint check failed: tool/resource definitions use about 1231 tokens (~153/item across 8 items; 8 tools + 0 resources), over budget; trim descriptions and params. See how to fix → Fail
- Usage-examples check failed: none of the tools include examples. See how to fix → Fail
Stability & Change Management70
- Stability observed for 21 of 30 days with no destabilising changes; credit accrues until the full window elapses.Partial
Tool Coverage84
- 100% of tools have a non-trivial description (not blank, and not just the tool's name).Pass
- 51% of tool parameters carry a description.Partial
Tool Safety100
- No prompt-injection markers were found in the server instructions, tool names or descriptions we captured.Pass
- We read all 8 captured tool definition(s), and no name or description among them implies an irreversible operation.Pass
- An AI judge read all 8 captured unit(s) of tool text and found none that tries to manipulate the model reading it.Pass
Capabilities60
- Spec-recency check failed: implements MCP spec 2025-06-18; the latest is 2026-07-28. See how to fix → Fail
Unverified: 1 category
A category scored 0 because we could not verify it: a data source with nothing on this package, evidence we could not reach, or a check we could not run. We only credit what we can confirm.
Every change we have recorded for this component, newest first. Security-relevant changes are always shown. ▲ marks a change for the better, ▼ a change for the worse; unmarked changes are neutral.
- 17 Oct 26 +1
No change was recorded against any check on this day. Stability & Change Management went from 63 to 67. That category is still filling its 30-day observation window: 19 days of observed history at the previous scan, 20 at this one. The score rises as the window fills, whether or not the server changes.
- 15 Oct 26 +1
No change was recorded against any check on this day. Stability & Change Management went from 57 to 60. That category is still filling its 30-day observation window: 17 days of observed history at the previous scan, 18 at this one. The score rises as the window fills, whether or not the server changes.
- 13 Oct 26 +1
No change was recorded against any check on this day. Stability & Change Management went from 50 to 53. That category is still filling its 30-day observation window: 15 days of observed history at the previous scan, 16 at this one. The score rises as the window fills, whether or not the server changes.
- 12 Oct 26 +1
No change was recorded against any check on this day. Stability & Change Management went from 43 to 50. That category is still filling its 30-day observation window: 13 days of observed history at the previous scan, 15 at this one. The score rises as the window fills, whether or not the server changes.
- 9 Oct 26 +6
- Stability: unverified → 0.40 ▲ functional
- 27 Sept 26 +39
- We updated how we score, so this day's move reflects our rubric, not a change to the server See what changed → functional
- 25 Sept 26 −4
- We updated how we score, so this day's move reflects our rubric, not a change to the server See what changed → functional
- 26 Aug 26 0
- We updated how we score, so this day's move reflects our rubric, not a change to the server See what changed → functional
Diagnostic detail from the automated scan of this channel: what the scanner observed at each step, so you can see exactly where a check passed or failed. It is informational only and never changes the trust score.
Captured 18 Oct 2026 · Analysed mcpb/https://github.com/aryawidjaja/marrow/releases/download/v0.8.0/marrow-mcp.mcpb@0.8.0
Provenance Verified
A signed build attestation was found and verified, binding this exact artifact to the source repository it claims to come from.
| Result | Verified |
|---|---|
| Ecosystem | mcpb |
| Reason | Verified |
| Discovered via | GitHub artifact attestations |
| Source repo | aryawidjaja/marrow |
| Certificate issuer | https://token.actions.githubusercontent.com |
| Certificate SAN | https://github.com/aryawidjaja/marrow/.github/workflows/release.yml@refs/tags/v0.8.0 |
| Rekor log index | 2405035141 |
| Predicate type | SLSA build provenance https://slsa.dev/provenance/v1 |
| Subject digest | sha256:ddcdf519d423dc1cc3e53c1c56d88e9e941ec02721c42c9f26e296a9e427f2c7 |
Background: How many MCP packages publish verified provenance →
The tools this component advertises to a client, with an estimated token cost for each. Expand a tool to see its parameters and schema. The per-tool counts are indicative and are not scored directly; the schema's total context footprint is one signal in Schema Quality & AI Usability. A tool's description is untrusted text the model reads on every call, which is what makes this list a security surface and not just an inventory: how tool poisoning works →
mem_areas ~45
The table of contents for this project's brain: its feature areas and how many memories each holds. Call this before mem_write so you file the memory into an area that already exists.
Input schema present but exposes no named parameters.
No output schema declared.
No examples provided.
mem_bootstrap ~82
Warm-start a session: announce it and get back what other sessions are doing plus the memories and decisions most relevant to your goal. Call this FIRST instead of re-scanning.
| Name | Type | Req | Description |
|---|---|---|---|
| by | string | – | – |
| goal | string | yes | – |
| max_tokens | integer | – | budget for relevant memories (default 1500) |
| project | string | – | – |
No output schema declared.
No examples provided.
mem_ingest ~54
Onboard an existing repo: list its knowledge docs (READMEs, docs/) so you can distill them into memory. Read each and save the durable decisions/facts with mem_write — distill, don't dump.
Input schema present but exposes no named parameters.
No output schema declared.
No examples provided.
mem_read ~26
Read a single memory by id, returned as markdown.
| Name | Type | Req | Description |
|---|---|---|---|
| id | string | yes | – |
No output schema declared.
No examples provided.
mem_recall ~255
Recall memories AND the ones connected to them (explicit links, shared topic/tag, related meaning) in one call — the related cluster, not just the matches. Records the retrieval so answers stay traceable.
| Name | Type | Req | Description |
|---|---|---|---|
| area | string | – | Feature area to favour (auth, billing, infra). BOOSTS that area to the top; it does not hide anything. |
| connect | integer | – | How many connected neighbours to return alongside the matches (default 8, 0 turns it off). |
| include_expired | boolean | – | – |
| kind | string | – | – |
| limit | integer | – | – |
| max_tokens | integer | – | – |
| min_confidence | number | – | – |
| project | string | – | Project scope. FILTERS: memories in other projects are never returned. |
| tag | string | – | Tag to favour. BOOSTS memories carrying it; it does not hide the ones that don't. |
| text | string | yes | – |
| topic | string | – | Topic to favour, e.g. `jwt-expiry`. BOOSTS matching topics; it does not hide anything, so a near-miss guess is safe. |
| weight | number | – | hybrid search weight: 0 keyword, 1 semantic |
No output schema declared.
No examples provided.
mem_search ~164
Hybrid keyword+semantic search over memories.
| Name | Type | Req | Description |
|---|---|---|---|
| include_expired | boolean | – | – |
| kind | string | – | – |
| limit | integer | – | – |
| max_tokens | integer | – | – |
| min_confidence | number | – | – |
| project | string | – | Project scope. FILTERS: memories in other projects are never returned. |
| tag | string | – | Tag to favour. BOOSTS memories carrying it; it does not hide the ones that don't. |
| text | string | yes | – |
| topic | string | – | Topic to favour, e.g. `jwt-expiry`. BOOSTS matching topics; it does not hide anything, so a near-miss guess is safe. |
| weight | number | – | hybrid search weight: 0 keyword, 1 semantic |
No output schema declared.
No examples provided.
mem_supersede ~120
Replace an existing memory with a new one, recording the lineage.
| Name | Type | Req | Description |
|---|---|---|---|
| area | string | – | Feature area this belongs to (auth, billing, infra). Call mem_areas and reuse an existing one. |
| body | string | yes | – |
| by | string | – | – |
| id | string | – | memory id being replaced (preferred; `old_id` remains accepted for compatibility) |
| kind | string | yes | – |
| old_id | string | – | deprecated alias for `id` |
| topic | string | – | Short label, not a sentence. |
No output schema declared.
No examples provided.
mem_write ~485
Write a new memory (a fact, a decision, or an entity). Rejects invalid writes with the reasons. FILE IT: pass `area` so the memory lands in the right part of the project's brain — call mem_areas first and REUSE an existing area rather than inventing a near-duplicate.
| Name | Type | Req | Description |
|---|---|---|---|
| anchor | object | – | ANCHOR IT IF IT IS ABOUT CODE. If this memory describes how a specific function/type behaves, name it here: Marrow fingerprints that symbol and flags this memory the moment the code changes, so the b… |
| area | string | – | The feature area this belongs to, e.g. `auth`, `billing`, `infra`. Call mem_areas and reuse one of the project's existing areas; only invent a new one if nothing fits. Leave it out if genuinely nothi… |
| body | string | yes | – |
| by | string | – | author for provenance |
| confidence | number | – | 0-1. Say so when you are NOT sure: an uncertain memory is useful, a confidently wrong one is not. Defaults to 1. |
| expires_at | string | – | RFC3339 timestamp after which this stops being true (e.g. a temporary workaround, a deadline). Marrow retires it automatically when it expires. Omit for anything durable. |
| kind | string | yes | – |
| model | string | – | YOUR model id, e.g. `claude-opus-4-8` or `gpt-5-codex`. Marrow cannot see this — the MCP handshake names your client, never the model — so pass it, or nobody will know which model formed this belief. |
| project | string | – | – |
| session | string | – | your session id, so the hive can attribute this |
| sources | array | – | Where this came from: file paths, URLs, docs you distilled it from. |
| tags | array | – | – |
| topic | string | – | Short label for what this is about, e.g. `jwt-expiry`. NOT a sentence — the detail goes in the body. Memories on the same topic supersede each other. |
No output schema declared.
No examples provided.
How do I install the Marrow MCP server?
Marrow ships as an MCPB bundle, a single file you download and open in an MCP host that supports MCPB bundles, such as Claude Desktop. The host reads the launch command from the bundle's own manifest, so there is no command to copy. The MCP registry declares no SHA-256 for this bundle, so there is no published digest to check the download against.
mcpb · marrow-mcp.mcpb
Download bundleThe MCP registry declares no SHA-256 for this bundle, so there is no published digest to check it against. Open it with an MCPB-capable host such as Claude Desktop.
What is the Marrow MCP server?
Marrow is an MCP server listed in the public MCP registry as io.github.aryawidjaja/marrow. Shared memory for parallel AI coding agents. Local, free, with rooms and file claims. This page covers its MCPB bundle (https://github.com/aryawidjaja/marrow/releases/download/v0.8.0/marrow-mcp.mcpb).
Is the Marrow MCP server safe to use?
Marrow scores 59 out of 100 on VerifyMCP. Its build provenance is signed and verified. That is a record of what we were able to check automatically, not an endorsement. The category breakdown on this page shows every signal behind the number, including the ones we could not confirm.
What tools does the Marrow MCP server expose?
Marrow exposes 8 tools: mem_write, mem_areas, mem_read, mem_search, mem_recall, and 3 more. Their descriptions and schemas cost roughly 1,231 tokens of context every time the server is loaded.
What licence is the Marrow MCP server under?
Marrow declares the AGPL-3.0-only licence, which is OSI-approved. That covers the source only, and says nothing about the cost of any service it calls.