[KEIAHNY]

PR Type

Enhancement

---

Description

• Add fallback AI models via EINFRA_FALLBACK_MODELS secret

• Increase AI timeout to 300 seconds

---

Diagram Walkthrough

flowchart LR
  workflow["PR Agent Workflow"] -- "fallback models from secret" --> fallback["config.fallback_models"]
  workflow -- "timeout set to 300s" --> timeout["config.ai_timeout"]

Loading

File Walkthrough

	Relevant files
Configuration changes	pr_agent.yml Configure fallback models and AI timeout for PR agent        .github/workflows/pr_agent.yml Replace empty config.fallback_models with secrets.EINFRA_FALLBACK_MODELS Add config.ai_timeout set to 300 seconds +2/-1

---

[github-actions]

PR Reviewer Guide 🔍

Here are some key observations to aid the review process:

⏱️ Estimated effort to review: 1 🔵⚪⚪⚪⚪

🧪 No relevant tests

🔒 No security concerns identified

⚡ Recommended focus areas for review Possible Issue config.fallback_models previously defaulted to the valid JSON value "[]" and is now sourced from ${{ secrets.EINFRA_FALLBACK_MODELS }}. If that secret is not configured or set to an empty value, the environment variable becomes an empty string, which pr-agent will attempt to parse as a JSON list and can fail on (unlike the explicit "[]" used before). Similarly, the secret must contain a valid JSON array (e.g. ["model-a", "model-b"]); a plain comma-separated value would not parse. This only manifests when the secret is missing, empty, or not in the expected JSON format. config.fallback_models: ${{ secrets.EINFRA_FALLBACK_MODELS }}

[github-actions]

PR Code Suggestions ✨

Explore these optional code suggestions:

Category	Suggestion	Impact
Possible issue	Quote fallback models secret to avoid null config If the EINFRA_FALLBACK_MODELS secret is unset or empty, this line evaluates to config.fallback_models: with no value, which YAML parses as null and can break the tool's config parsing. Wrap the expression in double quotes so it always renders as a string, matching the style of the other quoted config values. This ensures an empty secret gracefully resolves to an empty string instead of an invalid null value. .github/workflows/pr_agent.yml [29] - config.fallback_models: ${{ secrets.EINFRA_FALLBACK_MODELS }} + config.fallback_models: "${{ secrets.EINFRA_FALLBACK_MODELS }}" Suggestion importance[1-10]: 6 __ Why: If EINFRA_FALLBACK_MODELS is unset, the unquoted expression renders config.fallback_models: as YAML null, which can break config parsing. Quoting it matches the style of other quoted values like config.custom_model_max_tokens and safely yields an empty string.	Low
General	Default empty fallback models to valid JSON An empty EINFRA_FALLBACK_MODELS secret would set fallback models to an empty string, which may not be a valid JSON list (like the previous "[]" value). Guard the secret so the workflow falls back to a valid empty JSON array when the secret is missing, preventing config errors in the PR agent job. Use a conditional expression to supply "[]" as the default. .github/workflows/pr_agent.yml [29-30] - config.fallback_models: ${{ secrets.EINFRA_FALLBACK_MODELS }} + config.fallback_models: ${{ secrets.EINFRA_FALLBACK_MODELS != '' && secrets.EINFRA_FALLBACK_MODELS || '[]' }} config.ai_timeout: "300" Suggestion importance[1-10]: 5 __ Why: The suggestion correctly addresses that an empty secret would no longer produce the previous valid "[]" JSON default, potentially causing config errors. The conditional expression is valid GitHub Actions syntax, though it adds complexity compared to a simpler quoting or default-value approach.	Low

• Author self-review: I have reviewed the PR code suggestions, and addressed the relevant ones.