mcp@2.8.0

VERIFIED

Publisher claims: "Currents MCP server" Observed in sandbox: no network, touches nothing outside the project, writes no memory.

Registry npm (MCP server)Publisher currents · publishing for 596 daysInstalls 382KLast detonated 59 min ago
$claude mcp add mcp -- npx -y @currents/mcp
WatchReport an issue

Coverage

Started with a decoy CURRENTS_API_KEY; tool list observed. Tool calls with a real credential not tested.
Tested
server startup with a tagged decoy in CURRENTS_API_KEY · MCP initialize and tools/list · every tool description, for hidden instructions · where the decoy credential went · every file read, file written and host contacted by what ran
Not tested
what the tools do when called with a real credential
Needs
CURRENTS_API_KEY — give it to the server where it runs, through vet-skills run --secrets, so the value never enters the server's environment. Never paste a credential into a chat, a ticket or a report.

Before you give it your token

It will ask for
API_KEY_LINECURRENTS_API_KEYConfiguration, not credentials: CURRENTS_API_URL, CURRENTS_MCP_SURFACE, MCP_PATH
Where it talks to
currents.dev (its own)Also playwright.dev — not the publisher's domain; check what for before connecting.
What we saw when it ran
no credential files read · no network
What we could not test
what the tools do when called with a real credential

Connect it with a scoped credential, under the guard, and read the third-party hosts first. The code names playwright.dev as well as its own vendor. The guard injects API_KEY_LINE, CURRENTS_API_KEY only into requests to currents.dev (and its subdomains) and refuses the rest.

$vet-skills run --secrets API_KEY_LINE=currents.dev,CURRENTS_API_KEY=currents.dev -- npx -y @currents/mcp

Never paste a credential into a chat, a ticket or a report, including ours. How the guard holds it.

What changed since 2.7.0

vet-skills diff mcp 2.7.0 2.8.0

This version reaches for something 2.7.0 did not: 1 new tool (currents-list-test-issues) · 6 rewritten tool descriptors (currents-create-evidence-links, currents-create-webhook, currents-get-spec-instance, currents-get-test-evidence, …). Compared from the two recorded runs, not from source.

New tools
currents-list-test-issues
Rewritten tool descriptors
currents-create-evidence-links, currents-create-webhook, currents-get-spec-instance, currents-get-test-evidence, currents-get-webhook, currents-update-webhook

The guard can hold an update like this until someone approves it (runtime.on_change: hold). A change that only shows with a real credential, or on a path the run did not reach, is not in a trace and so not in this comparison.

Is [email protected] safe to install?

KillerSkills ran it in the KillerSkills honeypot sandbox (2 scenarios) on 2026-11-20 and scored it 87/100 (Verified). It did what it claims and nothing else we could detect: no undeclared network egress, no credential reads, no persistent instruction writes. Summary: no network, touches nothing outside the project, writes no memory.

Behavior trace

Honeypot run · 47 s · 2 scenarios · 0 deviations
t+CategoryObserved action
0.0sprocessno install scripts in bundle
7.0sprocessdependency install exit 0
27sprocessrelaunched with a decoy CURRENTS_API_KEY (tagged; nothing real)
47sclaimed44 tools with a decoy CURRENTS_API_KEY: currents-list-actions, currents-create-action, currents-get-action, currents-update-action, currents-delete-action, currents-enable-action, currents-disable-action, currents-list-affected-tests, currents-get-affected-test-executions, currents-get-action-executions, currents-get-projects, currents-get-project, currents-get-project-insights, currents-list-pull-requests, currents-list-project-terms, currents-create-jira-issue, currents-link-jira-issue, currents-list-jira-projects, currents-list-jira-issue-types, currents-get-runs, currents-get-run-details, currents-find-run, currents-cancel-run, currents-reset-run, currents-delete-run
47stools72d22b524383d56e
47stools59c988db04008b32
47stools30bd84f4a65320ec
47stoolsb19b6b59dc1c3d4c
47stools8a86aa3f791d2485
47stools235ed5339d7d0490
47stoolsc90f0adf73fd3b4d
47stools03c0afdeb7eeb1df
47stoolsdd95fb1ebe7b5c29
47stools95fd1a320599618b
47stools3a77371296a6fdbd
47stools8b6c36a7c013270d
47stoolsc67c73674bb42e5b
47stoolse1947b99f77e44b0
47stools8679b7ec6c35ea77
47stools62b8025b3608d6eb
47stoolsc253900bc9d135d6
47stoolsac137b37c039e412
47stools698bc3ad521f182b
47stoolsc5eaac200f7c994c
47stools071b70cb5f11c933
47stools5b1513e03343a0c7
47stoolsd034e52120a98089
47stools2134a1cdeb752c84
47stools66d1156c268df75c
47stools230cb47cbfbb7e38
47stools3a61d5830d8b8e7d
47stools144f9886d23654d5
47stools2d6ae5bb37149f0d
47stools24b25ba4a73393cd
47stools4f4fa53523dcc6be
47stoolscac045fdeccea0d6
47stools295dfea521ea1986
47stools3457c553a89b162f
47stoolse27fd7d8705750d1
47stools8668554b212312d9
47stoolsc75db15748385614
47stoolsc5e21922d69eb5cf
47stools4354a25b44854073
47stools1acfc84d281954bf
47stoolse80133ec94d5f599
47stools4a425ba964e4d02d
47stools83503ce1656303d3
47stoolsfa64f82a9e8d8a4b
47sprocessscript backend runs setup only
47sprocessno install scripts to re-run

Static findings

MEDIUM
File permission modification commandpackage.json:33 · File permission modification command — agent changes file ownership/permissions"build": "tsdown && chmod 755 dist/index.mjs dist/http.mjs",Fix: Request the narrowest permission that works; replace wildcards with the exact paths or hosts.
AST05
MEDIUM
npx invocation of non-well-known packagedist/logging-DuVICetC.cjs:120 · npx invocation of non-well-known package — may execute untrusted code…` with write access in the environment. In a project that has the package, run `npx currents`. Elsewhere, run `npx --package @currents/cmd currents`: `npx currents` alone run…Fix: Pin versions and sources; vendor anything fetched at install time.
AST06
MEDIUM
npx invocation of non-well-known packagedist/logging-DhrqXeHx.mjs:116 · npx invocation of non-well-known package — may execute untrusted code…only the tests you fixed, locally, when the repository allows it — for example `npx playwright test <spec> -g \"<test title>\"`. Do not report the local run to Currents.\n\nW…Fix: Pin versions and sources; vendor anything fetched at install time.
AST06
MEDIUM
npx invocation of non-well-known packagedist/logging-DuVICetC.cjs:139 · npx invocation of non-well-known package — may execute untrusted code…only the tests you fixed, locally, when the repository allows it — for example `npx playwright test <spec> -g \"<test title>\"`. Do not report the local run to Currents.\n\nW…Fix: Pin versions and sources; vendor anything fetched at install time.
AST06
MEDIUM
File type combined with upload actiondist/logging-DhrqXeHx.mjs:97 · File type combined with upload action — agent may exfiltrate local data files…identical. The tool returns the text; write it to `.currents-session/before-a11y.txt` yourself (`session start` makes `.currents-session/` ignored by git)\n- `browser_stop_tracing`, without a path\n\n### 3. Attach the files\n\nRun `current…Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.
AST07
MEDIUM
File type combined with upload actiondist/logging-DuVICetC.cjs:120 · File type combined with upload action — agent may exfiltrate local data files…identical. The tool returns the text; write it to `.currents-session/before-a11y.txt` yourself (`session start` makes `.currents-session/` ignored by git)\n- `browser_stop_tracing`, without a path\n\n### 3. Attach the files\n\nRun `current…Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.
AST07
MEDIUM
npx invocation of non-well-known packagedist/logging-DhrqXeHx.mjs:97 · npx invocation of non-well-known package — may execute untrusted code…` with write access in the environment. In a project that has the package, run `npx currents`. Elsewhere, run `npx --package @currents/cmd currents`: `npx currents` alone run…Fix: Pin versions and sources; vendor anything fetched at install time.
AST06
MEDIUM
File type combined with upload actiondist/logging-DuVICetC.cjs:128 · File type combined with upload action — agent may exfiltrate local data files…ait request.get('/api/cart/total?coupon=SAVE10');\n const body = await response.json();\n\n await testInfo.attach('evidence-cart-total.json', {\n body: JSON.stringify(body, null, 2),\n contentType: 'application/json',\n });\n\n expect(body…Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.
AST07
MEDIUM
File type combined with upload actiondist/logging-DhrqXeHx.mjs:105 · File type combined with upload action — agent may exfiltrate local data files…ait request.get('/api/cart/total?coupon=SAVE10');\n const body = await response.json();\n\n await testInfo.attach('evidence-cart-total.json', {\n body: JSON.stringify(body, null, 2),\n contentType: 'application/json',\n });\n\n expect(body…Fix: Remove the call, or declare the host in frontmatter and explain in the README what is sent and why.
AST07

Cost to the agent

Instructions loaded
≈39 tok
1 file · 13 in bundle
Tool list per prompt
≈16,728 tok
44 tools · 380 per tool
Handshake
4837 ms
launch to tools/list, stdio
Undocumented tools
0
description under 20 chars · avg 325

Tokens estimated at four characters each. The tool list is re-read by the agent on every turn, so it is the number to compare between servers that do the same job: ranked across the index.

Known vulnerabilities

Not checked against OSV yet; the weekly check has not reached this version.