Benchmark report
Blocked by a prerequisite
11m 49s elapsed
An external prerequisite prevented the documented task from completing. Benchmark again
15/ 100
Critical Gating- AInstall
- DAuth
- DExecute
- CDocs
- ACurrent
Mandatory egress proxy returned 403 on CONNECT for every invideo host, and the three approval requests came back 'skipped' and 'timed_out' twice.
Where the time went
8:00 on the owner · 3:49 agent time- Starting0:01
- Research0:26
- Attempt2:10
- Verify1:12
Judge reason
No invideo product use occurred. Every HTTPS attempt was refused by the mandatory egress proxy at the CONNECT stage: evidence/network-blocker-raw.txt shows '> CONNECT mcp.invideo.io:443' -> '< HTTP/1.1 403 Forbidden' -> 'curl: (56) CONNECT tunnel failed, response 403', with identical 403s for invideo.io/agents.md and /auth.md; WebFetch returned 'proxy refused the connection'. The final JSON-RPC tools/list POST to https://mcp.invideo.io/mcp returned http_code 000 (transcript idx 52-53). No tools/list response, no 'Morning Coffee Test' project id or editor URL, no generated image asset, and no timeline read-back exist. Setup completed before the block is credited: proxy configuration inspected and left unmodified, curl/python3/WebFetch availability confirmed, local .claude/.codex config searched for a pre-provisioned invideo MCP entry or cached credential (none found), and the block localized to proxy CONNECT rather than DNS/TLS. Egress approval was requested for invideo.io (outcome 'skipped') and twice for mcp.invideo.io (both 'timed_out'). The agent invoked no mocks, fabricated no ids, and did not call benchmark-success; /workspace/home/benchmark-first-success.json does not exist.
Attempt record
- 4:09Asked a TPC admin to allow mcp.invideo.iono approval within the window
- 10:37Blocked by a prerequisiteNetwork egress to mcp.invideo.io (and invideo.io) was never approved; the mandatory proxy returns HTTP 403 Forbidden on CONNECT for every required host, so no JSON-RPC request could reach the invideo MCP server.
Attempt artifacts saved.