fix(settings): restore missing large PR handling config for pr_description

[Hank076]

User description

Problem

Related #2230

Running python -m pr_agent.cli --pr_url="..." describe fails immediately with:

WARNING | pr_agent.algo.pr_processing:retry_with_fallback_models:333 - Failed to generate prediction with

All fallback models exhaust within a few milliseconds — no API call is
ever made.

Root Cause

pr_description.py line 208 reads:

large_pr_handling = get_settings().pr_description.enable_large_pr_handling and ...

The keys enable_large_pr_handling, max_ai_calls, and async_ai_calls
were removed from the [pr_description] section of
settings/configuration.toml. When the key is missing, get_settings()
raises an exception inside _prepare_prediction(), before get_pr_diff
is reached. retry_with_fallback_models() catches any Exception and
logs it as a prediction failure, silently exhausting every model in the
fallback list.

This is why the failure is /describe-specific: other tools do not
access pr_description.enable_large_pr_handling.

Fix

Restore the three missing settings under [pr_description] in
settings/configuration.toml:

# large pr mode
enable_large_pr_handling=true
max_ai_calls=4
async_ai_calls=true

How to Reproduce

Use the latest Docker image and run:

python -m pr_agent.cli --pr_url="<any_valid_pr_url>" describe

Observe all models fail instantly with no get_pr_diff log output,
while review and improve on the same PR work fine.


___

### **PR Type**
Bug fix


___

### **Description**
- Restore missing large PR handling configuration keys

- Fixes describe command failing with all models exhausted

- Re-adds enable_large_pr_handling, max_ai_calls, async_ai_calls settings


___

### Diagram Walkthrough


```mermaid
flowchart LR
A["Missing config keys<br/>in configuration.toml"] -->|causes| B["Exception in<br/>_prepare_prediction"]
B -->|caught by| C["retry_with_fallback_models"]
C -->|silently exhausts| D["All models fail<br/>instantly"]
E["Restore three keys:<br/>enable_large_pr_handling<br/>max_ai_calls<br/>async_ai_calls"] -->|fixes| A

File Walkthrough

	Relevant files
Bug fix	configuration.toml Add missing large PR handling configuration keys                  pr_agent/settings/configuration.toml Added enable_large_pr_handling=true configuration key Added max_ai_calls=4 configuration key Added async_ai_calls=true configuration key Placed under [pr_description] section with descriptive comment +4/-0

---

[qodo-free-for-open-source-projects]

PR Compliance Guide 🔍

Below is a summary of compliance checks for this PR:

Security Compliance
🟢	No security concerns identified No security vulnerabilities detected by AI analysis. Human verification advised for critical code.
Ticket Compliance
🟡	🎫 #2230 🟢 Restore missing [pr_description] configuration keys required by pr_description.py (enable_large_pr_handling, max_ai_calls, async_ai_calls) so get_settings() does not raise when preparing the prediction. ⚪ Fix the describe command failure in the latest Docker image where all fallback models are exhausted immediately without making an API call.
Codebase Duplication Compliance
⚪	Codebase context is not defined Follow the guide to enable codebase context checks.
Custom Compliance
🟢	Consistent Naming Conventions Objective: All new variables, functions, and classes must follow the project's established naming standards Status: Passed
	No Dead or Commented-Out Code Objective: Keep the codebase clean by ensuring all submitted code is active and necessary Status: Passed
	Robust Error Handling Objective: Ensure potential errors and edge cases are anticipated and handled gracefully throughout the code Status: Passed
	Single Responsibility for Functions Objective: Each function should have a single, well-defined responsibility Status: Passed
	When relevant, utilize early return Objective: In a code snippet containing multiple logic conditions (such as 'if-else'), prefer an early return on edge cases than deep nesting Status: Passed
Update

Compliance status legend

🟢 - Fully Compliant
🟡 - Partial Compliant
🔴 - Not Compliant
⚪ - Requires Further Human Verification
🏷️ - Compliance label

[qodo-free-for-open-source-projects]

PR Code Suggestions ✨

Explore these optional code suggestions:

Category	Suggestion	Impact
Possible issue	Ensure settings are in correct section Place the new large PR handling settings under a [pr_description] section header to ensure they are correctly scoped and applied to the intended tool. pr_agent/settings/configuration.toml [113-119] #custom_labels = ['Bug fix', 'Tests', 'Bug fix with tests', 'Enhancement', 'Documentation', 'Other'] + +[pr_description] # large pr mode 💎 enable_large_pr_handling=true max_ai_calls=4 async_ai_calls=true [pr_questions] # /ask # Apply / Chat Suggestion importance[1-10]: 9 __ Why: The suggestion correctly identifies a critical issue where new configuration options might be applied to the wrong section, leading to a misconfiguration, which is a functional bug.	High
Update

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

[drew-simmons]

I think inline_file_summary = false is missing too.

https://github.com/qodo-ai/pr-agent/blob/v0.31/pr_agent/settings/configuration.toml#L122

[naorpeled]

Hey @Hank076,
thanks for this, you rock 😎 🙏

Could you please address the comments by @drew-simmons and me?
Afterwards, I'll gladly merge this.

[PeterDaveHello]

We need this with a new release!

[Hank076]

Hey @Hank076, thanks for this, you rock 😎 🙏

Could you please address the comments by @drew-simmons and me? Afterwards, I'll gladly merge this.

Thanks for the review and feedback!
I'll take a look at the comments from you and @drew-simmons and update the PR.

[qodo-free-for-open-source-projects]

Code Review by Qodo

🐞 Bugs (2) 📘 Rule violations (0) 📎 Requirement gaps (0)

1. inline summary table mode missing 🐞 Bug ✓ Correctness

Description

inline_file_summary is documented in the default config as supporting 'table', but the code
appears to treat it only as a boolean (truthy/falsy) flag; users setting 'table' will not get
distinct behavior and may be misled about supported output formats.

Code

pr_agent/settings/configuration.toml[110]

+inline_file_summary=false # false, true, 'table'

Evidence

The default configuration explicitly advertises a 'table' mode, but the runtime logic only checks
the setting’s truthiness to decide whether to inline the walkthrough into the PR body (vs. using
file comments). No other implementation of a 'table' mode is referenced.

pr_agent/settings/configuration.toml[107-117]
pr_agent/tools/pr_description.py[125-132]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
`inline_file_summary` is documented as supporting values `false`, `true`, and `'table'`, but the code currently uses it only in a boolean truthiness check. This makes `'table'` behave the same as `true` and violates the documented contract.
### Issue Context
The setting is used to decide whether the changes walkthrough is appended to the PR body vs. published as file comments. There is no code path that detects `'table'` and formats output differently.
### Fix Focus Areas
- pr_agent/settings/configuration.toml[107-117]
- pr_agent/tools/pr_description.py[125-132]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

2. Describe docs miss new knobs 🐞 Bug ✓ Correctness

Description

max_ai_calls and async_ai_calls are now present in the default [pr_description] config and are
used by large-PR handling, but the describe tool documentation does not mention them, reducing
discoverability and making tuning large-PR behavior harder.

Code

pr_agent/settings/configuration.toml[R115-117]

+enable_large_pr_handling=true
+max_ai_calls=4
+async_ai_calls=true

Evidence

The restored settings directly affect large-PR behavior: max_ai_calls controls iteration count for
chunked diff generation, and async_ai_calls controls whether multiple model calls are
parallelized. The describe tool docs list enable_large_pr_handling but do not include these
related tuning parameters.

pr_agent/settings/configuration.toml[113-117]
pr_agent/algo/pr_processing.py[260-266]
pr_agent/tools/pr_description.py[244-263]
docs/docs/tools/describe.md[132-142]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
The default configuration exposes large-PR tuning parameters (`max_ai_calls`, `async_ai_calls`) that are used in runtime logic, but the describe tool documentation does not mention them.
### Issue Context
Users rely on docs to discover configuration options. These parameters control iteration count and concurrency for large PR handling.
### Fix Focus Areas
- pr_agent/settings/configuration.toml[113-117]
- docs/docs/tools/describe.md[110-151]

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

ⓘ The new review experience is currently in Beta. Learn more

[MontaEllis8]

✅ Code Review 通过

代码审查通过，建议合并。

[PeterDaveHello]

@naorpeled, is it possible we can go for it?

[naorpeled]

@naorpeled, is it possible we can go for it?

For sure 🙏

[Hank076]

Hi @naorpeled ,

The workflow is awaiting approval from a maintainer.
Could someone please approve and run the CI?

Thanks!

[naorpeled]

Thanks @Hank076!

[PeterDaveHello]

@naorpeled can we get a new release for this important fix? Many thanks!

[naorpeled]

@naorpeled can we get a new release for this important fix? Many thanks!

Hey Peter,
on it, it might take some time as the credentials reside within Qodo 🙏
(Soon this process will be easier and faster)

[PeterDaveHello]

Thank you for releasing the new version. I tried the latest v0.33, but it seems the same issue still exists for now. It might need some further investigation...