[cli-tools-test] agentic-workflows MCP: logs and audit tools always time out with context deadline exceeded after 60s

[github-actions]

Warning

threat detection engine error
The threat detection engine encountered an error and could not complete analysis. This is a tooling failure, not a security finding.

Details

The threat detection engine failed to produce results.

Review the workflow run logs for details.

Problem Description

Both the logs and audit tools in the agentic-workflows MCP server reliably fail after exactly ~60 seconds with Error [0]: context deadline exceeded. This happened on every invocation tested, including minimal/narrow queries that should return quickly. Meanwhile compile and status work correctly and return in well under 2 seconds.

Tool / Command

• Tools: logs, audit (agentic-workflows MCP server)
• Working tools for comparison: status, compile (both fast and correct)

Steps to Reproduce

1. Call logs with {"start_date":"-1d","count":3,"max_tokens":3000} — times out at 60021ms.
2. Call logs with {"workflow_name":"scout","count":1,"max_tokens":2000} — times out at 60019ms.
3. Call audit with {"run_id":30518297293} (the current, in-progress run) — times out at 60016ms.

All three attempts show the identical pattern in bridge logs:

[info] [agenticworkflows] MCP tools/call: status=200, elapsed=60019ms
[info] [agenticworkflows] MCP keepalive stopped
[info] [agenticworkflows] Tool call complete: total=60071ms
Error [0]: context deadline exceeded
[error] [agenticworkflows] Tool call error: Error [0]: context deadline exceeded

The underlying MCP tools/call response reports status=200 at exactly ~60s, suggesting a hard 60-second deadline/timeout is hit server-side (or in the gateway) for these two tools specifically, independent of the actual query scope or amount of data requested.

Expected Behavior

logs and audit should return within a reasonable time (docs suggest <10s for logs, <30s for most audits), or if the operation genuinely needs longer, the tool should support incremental/paginated output or a configurable timeout rather than unconditionally failing at 60s.

Actual Behavior

Every logs and audit call made during this test session (3 different parameter combinations across 2 tools) failed identically with context deadline exceeded at ~60s elapsed. No partial data, no actionable error message, no indication of what caused the delay (log download bottleneck, GitHub API pagination, backend processing, etc).

Environment

• Repository: github/gh-aw
• Run ID: 30518297293
• Tool: agentic-workflows MCP server (bridge to awmg-mcpg:8080)

Impact

• Severity: Critical — two of the three primary tools under test (logs, audit) are completely non-functional in this environment.
• Frequency: Always (3/3 for logs, 1/1 for audit tested)
• Workaround: None found. Reducing count, max_tokens, and narrowing by workflow_name/run_id did not help — all variants hit the same ~60s wall.

Additional Context

compile (multiple calls, both single- and multi-workflow) and status (full workflow enumeration, ~55 workflows) both completed in under 1.5s, so the MCP gateway itself is responsive — the issue appears isolated to the logs/audit code paths, possibly a slow upstream GitHub API log-download or artifact-processing step that exceeds an internal 60s deadline.

Generated by 🧪 Daily Cli Tools Tester · auto · 29.9 AIC · ⊞ 13.4K · ◷

• expires on Aug 5, 2026, 10:12 PM UTC-08:00