# rfp agent hit context limit mid questionnaire draft and failed to complete: how to fix it

An RFP agent hitting context limits mid questionnaire needs the same chunking discipline as any long document task. Process the questionnaire in sections, draft answers per section, and assemble at the end. Questionnaires are long by design, so plan for the limit from the start.

```text
rfp agent hit context limit mid questionnaire draft and failed to complete
```

## When to use this skill
- Questionnaire drafting dies on context limits
- Long questionnaires fail while short ones pass

## When not to use this skill
- Drafts are wrong rather than failing (a quality issue)
- The parser cant read the questionnaire (a parsing issue)

## Tool compatibility
Agent questionnaire drafting with LLM context limits.
Related area: model APIs and inference quotas.
Cloud UIs change labels over time; if a button moved, search settings for the closest match.

## Before you start
- Sign in to the tool and keep one tab open so sessions stay clean
- Copy the exact error text and note the time it happened
- Set aside 15 to 20 minutes; most of these resolve in a single pass

## Quick diagnosis
Make sure this is the right skill before you invest the 15 minutes.
- If you are dealing with questionnaire drafting dies on context limits, start at step 1 below.
- If instead drafts are wrong rather than failing (a quality issue), stop here; you need a different skill.
- If neither quite matches, compare your screen against the verbatim error block at the top; that block is exactly what this skill covers.

## Fix it in steps
### 1. Split the questionnaire into sections

```
Divide questions into logical groups that fit the context window.
```
Expected: sections sized for comfortable processing.

### 2. Draft answers per section

```
Process one section at a time with its relevant source material.
```
Expected: complete drafts for each section.

### 3. Assemble the full draft

```
Combine section drafts into the final questionnaire response.
```
Expected: a complete, coherent response.

### 4. Check for cross-section consistency

```
Review the assembled draft for contradictions between sections.
```
Expected: consistent answers throughout.

## Confirm each step worked
Run through this checklist before moving on. If a check fails, redo that step instead of pushing ahead.
- Step 1 (split the questionnaire into sections): Expected: sections sized for comfortable processing.
- Step 2 (draft answers per section): Expected: complete drafts for each section.
- Step 3 (assemble the full draft): Expected: a complete, coherent response.
- Step 4 (check for cross-section consistency): Expected: consistent answers throughout.

## Quick recap
1. Split the questionnaire into sections
2. Draft answers per section
3. Assemble the full draft
4. Check for cross-section consistency

## Still stuck?
Gather three things: the exact error text, the time it happened, and what changed just before it broke.
Check the vendor status page for an active incident. Then open a support ticket with those three things;
it cuts the back-and-forth in half. For agent runs, attach the last 50 lines of the agent log too,
since the loop or timeout signature in the log usually names the cause.

## Variant phrasings
### context limit questionnaire draft
Section it, draft per section, assemble. If this matches what you saw, the steps above apply as written.

### agent failed to complete questionnaire
Plan for the limit, questionnaires are always long. If this matches what you saw, the steps above apply as written.

## Why it happens
Questionnaires bundle hundreds of questions plus source material into one prompt, which overflows fixed context windows. Section-wise processing keeps each step inside the limit while the assembly pass restores coherence.

## Edge cases
- Keep related questions in the same section for consistency
- The assembly pass needs the full question list for numbering
- Save section drafts durably, long runs can still fail late

## Provenance

Resolved from the public thread: https://vectle.com/posts/pst_b497c1SHAbHEbDLuQzcfIA