Fix Research workflow - Critical failure (90% failure rate)

[github-actions]

Problem

The Research workflow is experiencing critical failures with a 90% failure rate (9/10 recent runs failed). The workflow has been essentially non-operational since 2026-01-08 (15 days ago).

Error Details

Last successful run: 2026-01-08
Recent failures: Multiple failures throughout January 2026
Latest failed run: §21078189533

Impact

• Research and knowledge work capabilities are severely limited
• Cannot perform automated research tasks
• Blocks workflows that depend on research functionality

Root Cause Analysis

Based on pattern analysis with similar failing workflows:

1. Likely Tavily MCP server issue - Similar to MCP Inspector and the recovered Daily News workflow
2. MCP Gateway startup failure - Expected to fail at "Start MCP gateway" step
3. Secret configuration - May need TAVILY_API_KEY or related credentials

Related Context

Pattern detected across Tavily-dependent workflows:

• ✅ Daily News: RECOVERED after TAVILY_API_KEY added (2026-01-22)
• ❌ MCP Inspector: FAILING (80% rate) - same MCP Gateway issue
• ❌ Research: FAILING (90% rate) - likely same root cause
• ⚠️ Scout: Status unknown (also uses Tavily)

Recommended Investigation Steps

1. Check workflow configuration:

   • Review .github/workflows/research.md frontmatter
   • Verify MCP server configuration (particularly Tavily)
   • Compare with Daily News (now working) and MCP Inspector (failing)

2. Verify secrets and environment:

   • Confirm TAVILY_API_KEY is set and accessible
   • Check if any other secrets are required
   • Review repository secrets configuration

3. Analyze recent failed runs:

   • Download artifacts from run 21078189533
   • Check MCP Gateway logs for specific error messages
   • Identify which step is failing (likely "Start MCP gateway")

4. Test workflow manually:

   • Trigger workflow run manually to reproduce issue
   • Add debug logging if needed
   • Compare behavior with working workflows (Smoke tests)

Success Criteria

• Research workflow runs successfully
• Success rate returns to >80% over next 5 runs
• Research functionality fully operational

Priority: P1 (High)

Research capabilities are critical for knowledge work and automated analysis. This workflow needs urgent attention to restore functionality.

References:

• Latest failed run §21078189533
• Workflow file

AI generated by Workflow Health Manager - Meta-Orchestrator

• expires on Jan 24, 2026, 2:59 AM UTC

[github-actions]

Status Update - 2026-01-24T02:51:00Z

Research Workflow Status

• Still failing: 90% failure rate persists (9/10 recent runs)
• Latest failure: §21078189533
• Days offline: 16 days (since 2026-01-08)
• Latest success: 1/10 runs succeeded (2026-01-10)

Pattern Analysis

Research workflow appears to have same root cause as MCP Inspector:

• Both use Tavily MCP server
• Both failing at MCP Gateway startup
• Both offline for 2+ weeks
• Daily News (also Tavily-dependent) recovered, but these two persist

Key Difference from Daily News

Daily News recovered immediately after TAVILY_API_KEY was added. Research and MCP Inspector did not. This suggests:

1. Additional configuration needed beyond TAVILY_API_KEY
2. Different MCP Gateway setup requirements
3. Possible workflow recompilation needed after secret was added

Recommended Actions

1. Analyze latest failure logs from run 21078189533
2. Compare frontmatter between:
   • Daily News (working) .github/workflows/daily-news.md
   • Research (failing) .github/workflows/research.md
   • MCP Inspector (failing) .github/workflows/mcp-inspector.md
3. Check if recompilation needed:

   gh aw compile .github/workflows/research.md

4. Test manually with workflow_dispatch trigger

Priority: P1 (High)

Research capabilities offline for 16 days. Needs urgent fix - likely same solution as MCP Inspector.

AI generated by Workflow Health Manager - Meta-Orchestrator

[github-actions]

This issue was automatically closed because it expired on 2026-01-24T02:59:48.757Z.