Is Synap Memory MCP server safe?
- Python filesystem write/delete
- Python filesystem read
- Python network egress
What to do: Nothing here argues against installing it. Grant the capabilities it lists only if you expect the tool to need them.
Synap Memory MCP server is an AI component analyzed by SkillTotal's deterministic static scanner. The scan found no malicious indicators, though 1 risky construct is reported for review. It can: filesystem read, filesystem write, mcp tools detected and network egress — capabilities are what the code can do, not a verdict on intent. Risk score 10/100 (low).
repo
Automated static-analysis result. It can contain false positives and false negatives, and is not a claim about the intent of Synap Memory MCP server's authors. Report a false positive.
Behavioral traits
How this component maps to the CSA agentic threat model. Descriptive — it never affects the risk score.
Findings (8)
A server is bound to all network interfaces (0.0.0.0), not just your own machine.
uvicorn.run(app, host="0.0.0.0", port=settings.port, timeout_keep_alive=75)
Why it matters: Without authentication, other hosts on the network can reach it.
Fix: Bind to 127.0.0.1 for local-only use, or require authentication and restrict access if remote exposure is intended.
The component reads files from disk.
const raw = JSON.parse(fs.readFileSync(credPath, 'utf-8')) as Record<string, string>;
text = this.#fs.readFileSync(this.path, 'utf8');
Why it matters: Usually legitimate, but worth confirming it can't be steered into reading sensitive files.
Fix: Confirm which files are read and that paths cannot be influenced by untrusted input to reach sensitive locations.
The component writes or deletes files on disk.
fs.writeFileSync(this.credentialsPath, JSON.stringify(creds, null, 2), { mode: 0o600 });this.#fs.writeFileSync(tmp, body, 'utf8');
try { this.#fs.unlinkSync(this.path); } catch { /* already gone */ }this.#fs.appendFileSync(this.path, `${JSON.stringify(record)}\n`, 'utf8');Why it matters: Usually legitimate, but worth confirming the paths can't be controlled by untrusted input.
Fix: Confirm which files are written/deleted and that paths cannot be influenced by untrusted input.
The component reads files from disk.
return json.loads(_CONTRACT_PATH.read_text(encoding="utf-8"))
with open(file_path, "rb") as fh:
Why it matters: Usually legitimate, but worth confirming it can't be steered into reading sensitive files.
Fix: Confirm which files are read and that paths cannot be influenced by untrusted input to reach sensitive locations.
The component writes or deletes files on disk.
(root / "AGENTS.md").write_text(_as_markdown(fixture.memories), encoding="utf-8")
with open(args.out, "w", encoding="utf-8") as handle:
with open(args.out, "w", encoding="utf-8") as handle:
Path(args.out).write_text(json.dumps(rows, indent=2))
shutil.rmtree(item)
Why it matters: Usually legitimate, but worth confirming the paths can't be controlled by untrusted input.
Fix: Confirm which files are written/deleted and that paths cannot be influenced by untrusted input.
The component makes outbound network requests.
const response = await sdk.fetch({const response = await ctx.sdk.fetch({fetch(args: SynapFetchArgs): Promise<SynapFetchResponseLike>;
const response = await sdk.fetch({fetch(args: SynapFetchArgs): Promise<SynapFetchResponseLike>;
* When true (default), a fresh recall() triggers sdk.fetch(search_query=[...])
const response = await this.sdk.fetch({const response = await sdk.fetch({fetch(args: SynapFetchArgs): Promise<SynapFetchResponseLike>;
const req = https.request(url, {const res = await fetch(url, {const response = await fetch(`${baseUrl}/api/v1/memories/create`, {* - ContextAssembledEvent on every fetch (any source)
await step('user.context.fetch()', async () => {const context = await client.user.context.fetch({await step('cross-scope fetch() produces a formatted prompt', async () => {const unified = await client.fetch({await client.conversation.context.fetch({ conversation_id: 'not-a-uuid' });* - **Namespaced, snake_case** (`client.user.context.fetch(...)`) is the
/** Options for the cross-scope `fetch()`. Snake_case, matching Python's kwargs.
/** @deprecated Use `client.user.context.fetch()`. Note it returns raw snake_case. */
/** @deprecated Use `client.customer.context.fetch()`. */
/** @deprecated Use `client.client.context.fetch()`. */
* @deprecated Use `client.user.context.fetch()` and read the typed collections.
* @deprecated Use `client.user.context.fetch()` with an empty `search_query`.
Why it matters: Usually legitimate, but confirm the destinations are expected and no sensitive data leaves.
Fix: Confirm the destination hosts are expected and that no sensitive data is sent off-host.
The component makes outbound network requests.
async with httpx.AsyncClient(
base_url=settings.synap_api_url, timeout=timeout
) as client:from .requests import RequestEnvelope, ResponseEnvelope
self._client = httpx.AsyncClient(
base_url=base_url,
timeout=httpx.Timeout(5.0), # Short timeout for telemetry
)timeout=httpx.Timeout(5.0), # Short timeout for telemetry
self._limits = httpx.Limits(
max_connections=20,
max_keepalive_connections=10,
keepalive_expiry=(
keepalive_expiry
if keepalive_expiry is not None
else self …self._client = httpx.AsyncClient(
base_url=self.base_url,
timeout=httpx.Timeout(
connect=self.timeouts.connect,
read=self.timeouts.read,
write=self.timeouts.write, …timeout=httpx.Timeout(
connect=self.timeouts.connect,
read=self.timeouts.read,
write=self.timeouts.write,
pool=self.timeouts.connect,
),Why it matters: Usually legitimate, but confirm the destinations are expected and no sensitive data leaves.
Fix: Confirm the destination hosts are expected and that no sensitive data is sent off-host.
An MCP tool surface (manifest or tool definitions) was found.
"""Call a MAF ``@tool``-decorated function, sync or async."""
- **Tools** (`@tool`): `create_synap_tools` returns `search_memory` / `store_memory`
`create_synap_tools` returns two Smolagents `@tool` functions — `search_memory` and
``@tool`` functions for agents that don't adopt ``MemoryManager``.
Provides ``search_memory`` and ``store_memory`` as Strands ``@tool`` functions
Returns ``[search_memory, store_memory]`` as Strands ``@tool`` functions
@mcp.tool(description=LOG_DESC)
@mcp.tool(description=STATUS_DESC)
@mcp.tool(description=RECALL_DESC)
@mcp.tool(description=LIST_DESC)
Why it matters: Just context — review which tools it offers and their permissions.
Fix: Review the declared MCP tools and their permissions.
Check your own component
Run the same evidence-backed scan on any MCP server, agent skill, or package.
Scan your own componentHow we determine this: deterministic static analysis (regex + AST), evidence-anchored, no code execution. Methodology →