Replace GPTR content generation with custom flow

[gregagi]

Summary

• replace GPTR-based blog post generation in BlogPostTitleSuggestion.generate_content with the existing PydanticAI content generation agent
• add validation + retry logic to reject placeholder language (for example: "insert image here") and incomplete/cut-off endings
• keep post-generation link insertion flow intact so link exchange backlinks are still injected for Pro projects
• tighten SEO content prompt guidance to avoid placeholder directives

Notes

• no new tests were added (repo guidance says avoid adding/running tests unless explicitly requested)
• syntax sanity check executed: uv run python -m py_compile core/models.py core/prompts.py

[coderabbitai]

Warning

Rate limit exceeded

@gregagi has exceeded the limit for the number of commits that can be reviewed per hour. Please wait 20 minutes and 24 seconds before requesting another review.

⌛ How to resolve this issue?

After the wait time has elapsed, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

We recommend that you space out your commits to avoid hitting the rate limit.

🚦 How do rate limits work?

CodeRabbit enforces hourly rate limits for each developer per organization.

Our paid plans have higher rate limits than the trial, open-source and free plans. In all cases, we re-allow further reviews after a brief timeout.

Please see our FAQ for further information.

📥 Commits

Reviewing files that changed from the base of the PR and between 4e1e270 and 52c4968.

📒 Files selected for processing (2)

• core/models.py
• core/prompts.py

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Post copyable unit tests in a comment
• Commit unit tests in branch feat/replace-gptr-custom-content-flow

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[code-review-doctor]

Looks good. Worth considering though. View full project report here.

[code-review-doctor]

Suggested change

	logger.warning(
	"[Generate Content Custom Flow] Generated content failed validation",
	title_suggestion_id=self.id,
	project_id=self.project.id,
	generation_attempt_number=generation_attempt_number,
	maximum_generation_attempts=maximum_generation_attempts,
	validation_error=validation_error,
	)
	logger.warning(
	"[Generate Content Custom Flow] Generated content failed validation",
	title_suggestion_id=self.id,
	project_id=self.project_id,
	generation_attempt_number=generation_attempt_number,
	maximum_generation_attempts=maximum_generation_attempts,
	validation_error=validation_error,
	)

[code-review-doctor]

self.project.id performs a database read when id is evaluated. You could take advantage of Django's caching of related fields by using self.project_id, which does not do a database read. Explained here.

[greptile-apps]

Greptile Summary

Replaced GPTR-based blog generation with custom PydanticAI agent flow, adding validation and retry logic to prevent placeholder language and incomplete content.

Major changes:

• Removed GPTResearcher dependency and run_gptr_synchronously utility import
• Added create_generate_blog_post_content_agent for structured content generation using existing PydanticAI agent
• Implemented contains_placeholder_language() to detect editorial placeholders (e.g., "insert image here", "[TODO]")
• Implemented has_incomplete_ending() to validate content ends with complete sentences
• Added retry loop (up to 3 attempts) with validation feedback to agent
• Improved get_internal_links() to deduplicate pages by URL
• Updated prompts to emphasize direct inline links instead of placeholder suggestions
• Preserved post-generation link insertion flow for Pro projects

Confidence Score: 4/5

• This PR is safe to merge with minor issues
• The refactoring is well-structured with comprehensive validation logic and proper error handling. The migration from GPTResearcher to PydanticAI is clean, but there's a minor regex issue and redundant validation check that should be addressed.
• Check core/models.py:866 for the regex pattern fix

Important Files Changed

Filename	Overview
core/models.py	Replaced GPTResearcher with custom PydanticAI flow, added validation/retry logic for placeholder detection and incomplete endings, improved link deduplication
core/prompts.py	Updated SEO content prompts to remove placeholder suggestions and emphasize direct inline links instead of editorial notes

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[Start: generate_content] --> B[Create PydanticAI Agent]
    B --> C[Build Generation Context]
    C --> D{Attempt Loop<br/>Max 3 attempts}
    D --> E[Build Prompt with<br/>Previous Error]
    E --> F[Run Agent Synchronously]
    F --> G[Get Generated Schema]
    G --> H{Validate Content}
    H -->|Empty| I[Set Error: Empty]
    H -->|Has Placeholders| J[Set Error: Placeholders]
    H -->|Incomplete Ending| K[Set Error: Incomplete]
    H -->|Valid| L[Return Schema]
    I --> M{More Attempts?}
    J --> M
    K --> M
    M -->|Yes| D
    M -->|No| N[Raise ValueError]
    L --> O[Create Blog Post]
    O --> P[Insert Links]
    P --> Q[Process Content]
    Q --> R[Update Title & Slug]
    R --> S[Save Blog Post]

Loading

_{Last reviewed commit: 52c4968}

[greptile-apps]

_{2 files reviewed, 2 comments}

_{Edit Code Review Agent Settings | Greptile}

[greptile-apps]

regex pattern escapes hyphen - unnecessarily

Suggested change

	if re.search(r"[:;,\-(\[]$", last_line) or last_line.endswith("..."):
	if re.search(r"[:;,\-\([]$", last_line) or last_line.endswith("..."):

[greptile-apps]

redundant empty content check after validation

The validate_generated_blog_post_content method already checks for empty content on line 881-882, so this check is redundant and will never be reached.