tavily-mcp

npm / sealed 2026-06-16 / No. 4232e1cf

Does tavily-mcp MCP work in Claude Code and Cursor? Throne executed tavily-mcp mcp in a single-use Firecracker microVM and replayed both recorded client behaviours. Compatibility verdict: fit. Is tavily-mcp mcp safe? Static security scan: 2 finding(s) to review under security ruleset v1. The full per-step results, security findings, scan date, and evidence hash are below.

re-scan this server gate it in your CI

> throne registry tavily-mcp sealed

receiptsealed evidence

scan id: 4232e1cffa5e482cb4e35b642a77ca0e
target: tavily-mcp
sealed at: 2026-06-16 18:57:03Z
evidence hash: sha256:0a58e3716b87075539da47029f62c1f01577cabe92fada8df13a5aaefb00b54d

01connectPASS903ms

initialize ok: server tavily-mcp 0.2.20, negotiated protocolVersion 2025-11-25, capabilities ['tools']

02discoverPASS5ms

supported: tools/list (5 tools); method not found (tolerated): resources/list, prompts/list

03validate_schemasPASS17ms

all 5 tool inputSchemas are valid JSON Schema

04smoke_test_toolsPASS140ms

called 5 of 5 tools (cap 10, strategy=serial): 4 ok, 1 returned tool-level errors (expected for synthesized args)

05error_handlingPASS5.2s

structured error responses, connection survived, wrong_type_args: tavily_search:accepted (returned result); tavily_extract:accepted (returned result); tavily_crawl:accepted (returned result); tavily_map:accepted (returned result); tavily_research:accepted (returned result); unknown_method: error(-32601); invalid_id: no response (silent), connection alive; oversized_input: error(-32601); recovery: accepted (returned result); cancellation: error(-32601); old_version: accepted old version 2024-10-07, negotiated 2024-10-07, server may have stale protocol handling

06streamingNOT_APPLICABLE0ms

no streaming-capable tool declared by this server

07resource_lifecycleNOT_APPLICABLE1ms

resources capability not declared

08concurrent_callsPASS20ms

ladder 1 overlapping tavily_search calls: max_observed_stable_concurrency=1; all ids answered exactly once at every level

09reconnectPASS899ms

transport closed and relaunched; re-handshake ok (protocolVersion 2025-11-25, first session was 2025-11-25)

01connectPASS1.8s

initialize ok: server tavily-mcp 0.2.20, negotiated protocolVersion 2025-11-25, capabilities ['tools'] [request_strategy=pipelined, connection_count=2, second connection established]

02discoverPASS6ms

supported: tools/list (5 tools); method not found (tolerated): resources/list, prompts/list; notes: connection 2 confirms 5 tools

03validate_schemasPASS17ms

all 5 tool inputSchemas are valid JSON Schema

04smoke_test_toolsPASS87ms

called 5 of 5 tools (cap 10, strategy=pipelined): 4 ok, 1 returned tool-level errors (expected for synthesized args)

05error_handlingPASS5.1s

06streamingNOT_APPLICABLE0ms

no streaming-capable tool declared by this server

07resource_lifecycleNOT_APPLICABLE1ms

resources capability not declared

08concurrent_callsPASS143ms

ladder 1/2/4/6 overlapping tavily_search calls: max_observed_stable_concurrency=6; all ids answered exactly once at every level

09reconnectPASS1.9s

transport closed and relaunched; re-handshake ok (protocolVersion 2025-11-25, first session was 2025-11-25)

chatgpt desktop calibratingemulation profile pending real-traffic captureCOMING SOON

SECURITY: REVIEW / 2 finding(s) / review recommended

LOWTHR-INSTALL-03 / Install-time script execution (npm lifecycle)

"prepare" runs on git-dependency installs and local dev (not registry installs): 'npm run build'

package/package.json

LOWTHR-NET-05 / Hardcoded outbound endpoints

2 non-local endpoint host(s) referenced in code, verify each is expected for this server's purpose: api.tavily.com (package/build/index.js:21), docs.tavily.com (package/build/index.js:28)

package/build/index.js:21

VERDICT: FIT14 of 18 steps exercised across 2 clientsSANDBOXED RUN, submitted server executed in a disposable microVM, compatibility: 0 fail / 0 warn (14 of 18 steps exercised across 2 clients) / security: review, 2 finding(s), 0 highsealed by THRONE / No. 4232e1cf / 2026-06-16

scopeattestation tuple

target: tavily-mcp npm
engine: sandboxed
claude code: calibration: partial recorded=9,spec=0,assumed=9
cursor: calibration: partial recorded=11,spec=0,assumed=7
chatgpt desktop: calibration: unavailable recorded=0,spec=2,assumed=16
steps exercised: 14 of 18
test suite: v1.0.0
security ruleset: v1.0.0
sealed at: 2026-06-16 18:57:03Z
evidence hash: sha256:0a58e3716b87075539da47029f62c1f01577cabe92fada8df13a5aaefb00b54d
valid until: 2026-09-14

wear the crown

paste this in your README. it renders the live verdict and links back to this record. if a release ever breaks the verdict, the badge says so on its own.

[![THRONE: FIT](https://api.usethrone.dev/api/badge/tavily-mcp)](https://usethrone.dev/server/tavily-mcp)

executed in a disposable microVM, created for this scan and destroyed after it. nothing outlives a run.

maintainer of this server? challenge this record: hello@usethrone.dev. tell us what we got wrong and we re-run it in the open.

this page renders the stored record of a real run. nothing on it is asserted without the execution that proved it.