[cli-tools-test] agenticworkflows MCP `logs` tool always times out with context deadline exceeded

[github-actions]

Problem Description

The logs tool of the agentic-workflows MCP server consistently fails with context deadline exceeded after exactly ~60 seconds, for every parameter combination tried. This makes the logs command completely unusable via MCP.

Command/Tool

• Tool: logs
• Command (CLI bridge form):
  • agenticworkflows logs --start-date "-1d" --count 3 --max-tokens 3000
  • agenticworkflows logs --count 1 --max-tokens 1000

Steps to Reproduce

1. Call the agenticworkflows MCP server logs tool with start_date: "-1d", count: 3, max_tokens: 3000.
2. Observe the call hangs, sending keepalive pings every ~10s.
3. After 60018ms the MCP tools/call returns status=200 but the CLI bridge then reports Error [0]: context deadline exceeded.
4. Repeat with minimal params (count: 1, max_tokens: 1000, no date filter) — same failure, same ~60s timeout.

Expected Behavior

The logs tool should return a bounded set of recent workflow run logs within a reasonable time (target <10s per the testing spec), or return a clear, fast error if the operation cannot complete.

Actual Behavior

Every invocation (2/2 attempts, with different parameter sets) hangs for exactly 60 seconds and then fails with a generic context deadline exceeded error, with no partial results, no indication of what step is slow, and no way to work around it via smaller counts/params.

Environment

• Repository: github/gh-aw
• Run ID: 30333543112
• 264 workflows discovered via status, all reporting compiled: Yes

Impact

• Severity: High — blocks the logs tool and any downstream flow (audit-from-logs, log-based triage) depending on it via MCP.
• Frequency: Always (2/2 reproductions with different params).
• Workaround: None found. audit --run-id <id> works directly when the run ID is already known (e.g. from GitHub context), so audit is usable standalone, but discovering run IDs via logs is not possible.

Logs/Diagnostics

[info] [agenticworkflows] MCP tools/call: tool=logs, args={"start_date":"-1d","count":3,"max_tokens":3000}
[info] [agenticworkflows] MCP keepalive ping: id=1000, elapsed=7ms
[info] [agenticworkflows] MCP keepalive ping: id=1004, elapsed=8ms
[info] [agenticworkflows] MCP tools/call: status=200, elapsed=60020ms
[info] [agenticworkflows] MCP keepalive stopped
[info] [agenticworkflows] Tool call complete: total=60058ms
Error [0]: context deadline exceeded
[error] [agenticworkflows] Tool call error: Error [0]: context deadline exceeded

Second reproduction (minimal params, no date filter):

[info] [agenticworkflows] Calling tool 'logs' with args: {"count":1,"max_tokens":1000}
... (same keepalive pattern) ...
[info] [agenticworkflows] MCP tools/call: status=200, elapsed=60017ms
Error [0]: context deadline exceeded

Additional Context

• Other MCP tools work correctly: status (264 workflows enumerated, fast), compile (individual + multi-workflow compiles succeed in <1s, error handling for nonexistent workflows is clean and actionable), and audit (rich report with API quota/network/timeline data returned in <350ms, clean error messages for invalid run IDs).
• Minor separate observation: compile's CLI bridge only accepts --workflows (array) not --workflow-name; passing --workflow-name gives a clear self-correcting error, so not filed separately.

Generated by 🧪 Daily Cli Tools Tester · sonnet50 · 30.6 AIC · ⌖ 7.97 AIC · ⊞ 13K · ◷

• expires on Aug 3, 2026, 10:13 PM UTC-08:00