context limit error mid large-file remediation agent run

context limit error mid large-file remediation agent run - how to fix it

TL;DR

Exclude the large files from the working set and patch by slice: skip bundles and generated files entirely, load only the code around each violation, and process violations one at a time with progress on disk. One line of why: a single 2MB file can eat the whole context, and the agent never needed the whole file.

The error, verbatim

RemediationAgentError: context limit exceeded
    file: generated/api-client.js (2.1MB, 41000 lines)
    violations_in_file: 3 (all in the header comment block)

Fix it step by step

Step 1: Reproduce the overflow

node agent/remediate.js --file generated/api-client.js | rg -i 'context|limit' | head -3

Expected: The run dies loading the giant file.

Step 2: Confirm the violations are local

npx @axe-core/cli https://example.com/ --save axe-big.json | tail -2

Expected: The violations are in a small region of the file.

Step 3: Add slice-based patching

rg -n 'slice|window|chunk' agent/remediate.js | head -10

Expected: Add: load 200-line windows around each violation, exclude generated files by pattern.

Step 4: Re-run remediation

node agent/remediate.js | tail -4

Expected: Agent patches the slices and completes within budget.

Step 5: Add a regression probe

node agent/run-audit.js --smoke | tail -3

Expected: Smoke run passes; schedule it so the breakdown is caught if it ever regresses.

When to use this skill

• You run an agent that scans UIs for accessibility and it hits this breakdown
• The agent's scan loop stalls, crashes, or loops on this exact failure
• You are hardening an audit agent's error handling for production scans

When NOT to use this skill

• A human runs the scan manually and it works, this is agent-harness failure handling
• The scan completes and only reports violations, use the rule-specific skills

Compatibility

Remediation agent on any LLM backend. Slicing is harness logic around the violation locations. Pin the tool version in the lockfile so scans stay reproducible across machines.

Variant phrasings

agent context limit large file

Same overflow, same slicing.

remediation OOM generated file

Practitioner phrasing.

the breakdown hits other routes too

Agent failure modes are systemic; apply the hardening to every route the agent covers, not just the one that failed.

Why it happens

Generated files (API clients, bundles, vendored code) are huge and occasionally contain violations, and agents that load whole files into context die on them. The violations are always localizable: axe reports selectors or line hints, and the patch needs only the surrounding slice. The harness should exclude generated files by pattern by default (they should be fixed at the generator, not patched), and for legitimately large source files, window the context around each violation. Agent breakdowns are systemic: the same failure mode will hit every route, page, or run the agent touches. Harden the harness once (timeouts, loop detection, verification gates) instead of patching per page, and keep breakdown telemetry separate from violation counts.

Edge cases

• Generated files get fixed at the generator or config, not by patching the output.
• 200-line windows cover nearly every violation's context, tune only if patches keep missing.
• Track context usage per file and skip-with-report above a threshold, never crash the run.
• Log breakdowns separately from violations in agent telemetry; mixing them hides whether the agent itself is getting more reliable.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_kR05f1zz4rHrVZBTwt5IsQ