fix(agent): bound generated research queries

[iamjr15]

Why

Production QA of the canonical Markdown research PDF flow exposed a mismatch between accepted research topics and generated provider queries. A valid detailed topic could exceed the narrower generated-query schema once an angle suffix was appended, causing research_deep to fail validation and the agent to fall back to manual research.

What changed

• align generated research queries with the provider's 2,000-character query boundary
• preserve every distinct research angle when a long topic must be shortened
• apply the same safe composition to fan-out goals and entity queries
• clarify that tool-generated topics should remain concise and focused

Architecture and migration impact

No database, deployment-topology, or public API changes. This is an internal agent workflow validation fix.

Verification

• pnpm lint
• pnpm typecheck
• pnpm turbo build --force
• pnpm deadcode
• pnpm architecture:check
• pnpm turbo skills:build
• targeted 2,000-character deep-research and fan-out query boundary exercise, including distinct-angle and suffix-preservation assertions
• production natural-language research flow will be rerun after merge/deploy

The local machine reports the existing Node engine warning (repo pins 24.18.0; local Node is 26.4.0); all checks completed successfully.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

1 Skipped Deployment

Project	Deployment	Actions	Updated (UTC)
cheatcode	Skipped		Aug 5, 2026 5:17am