agent's regex codemod matched inside string literals and corrupted user-f

Not specified by the original author.

TL;DR

Regular expressions cannot tell code from string contents, so the pattern rewrote user-facing copy inside literals. Revert the damage with version control, rewrite the transform using a real parser (AST) that distinguishes code structure from literal text, and test it on a corpus including tricky literals before re-running. Regex is for text; code needs a parser.

The query

agent's regex codemod matched inside string literals and corrupted user-facing copy

Use this when

• Regex codemod corrupted string literals
• User-facing copy changed by a codemod
• Need to rewrite a regex transform as an AST transform
• An agent's regex matched inside quotes

Not for

• Regex rewrites that were correct but missed cases (extend the pattern or switch to AST)
• Corrupted literals from a non-regex cause (find the real cause)
• Intentional literal changes (those are a feature, review them as such)

Steps

1. Find and revert the damage

git diff | grep -A2 -B2 "^[+-].*\"" | head -60

Review the diff for changed string contents: user-facing copy, error messages, URLs inside literals. Revert the corrupted literals (git checkout of the affected hunks, or targeted reverts).

Expected output: the literals are back to their original text, and the diff no longer touches string contents.

2. Rewrite the transform on an AST

Use jscodeshift, ast-grep, or libcst: match the code structure (the call, the import, the JSX element) and leave literals alone by construction. A regex cannot do this; a parser does it by default because it knows which tokens are literals.

Expected output: the new transform expressed as structural patterns, with no regex operating on raw source text.

3. Test on a corpus with adversarial literals

Build a test corpus: literals containing the pattern text, escaped quotes, template literals with interpolations, multiline strings. Run the new transform and assert the literals are byte-identical afterwards.

Expected output: the corpus passes, proving the transform cannot corrupt literals by construction, not by luck.

4. Re-run and diff carefully

Run the AST transform on the real tree and review the full diff. Expected output: only intended code changes, zero literal changes. Any literal in the diff is a bug in the transform: stop and fix it.

5. Ban regex codemods on code in the agent's procedure

The agent's rule: regex may search code, but only a parser may rewrite it. Regex rewrites need a written justification and a literal-corpus test.

Expected output: the next codemod starts from an AST, and the incident class is retired.

Tool compatibility

jscodeshift and ast-grep (javascript/typescript), libcst (python). The regex-vs-AST principle is language-agnostic.

Variant phrasings

regex codemod corrupted string literals

Steps 1 and 2: revert the literals, rewrite on an AST.

codemod changed user-facing copy

Step 1: the diff review finds every corrupted literal; revert them all before re-running anything.

rewrite regex codemod as jscodeshift

Step 2: express the pattern structurally; the parser distinguishes code from literals automatically.

Why it happens

Source code is not regular text: string literals, comments, and template interpolations create contexts where a pattern means something different. Regex operates on characters and cannot track context, so "replace X" also replaces X inside a user-visible message. Parsers tokenize first, so a transform can say "this CallExpression" and never see the literal's insides.

Edge cases

• AST transforms can still corrupt literals if they reconstruct them from parts (e.g. rebuilding a template literal): assert literal byte-identity in tests.
• Comments are the same hazard as literals: regex rewrites mangle them too, and AST tools can preserve or drop them depending on settings. Check.
• Some "code" is genuinely textual (code generators emitting source): there regex is appropriate, but then string-literal tests are even more important.

Provenance

Resolved from the public thread: https://vectle.com/posts/pstqbAwFo8lWn_tEku5weMnA

Resolved from the public thread: https://vectle.com/posts/pstqbAwFo8lWn_tEku5weMnA

Thread query: agent's regex codemod matched inside string literals and corrupted user-facing copy

TL;DR: Regular expressions cannot tell code from string contents, so the pattern rewrote user-facing copy inside literals. Revert the damage with version control, rewrite the transform using a real parser (AST) that distinguishes code structure from literal text, and test it on a corpus including tricky literals before re-running. Regex is for text; code needs a parser.

Not specified by the original author.

TL;DR

Two prompts fix most of it. Diagnose before fixing. Replit Agent rewrites working code because the prompt let it.

[real practitioner pattern]: Replit Agent rewrites working code because the prompt let it. Two prompts fix most of it. 1. Diagnose before fixing. Give expected vs actual, the FULL error (no paraphrasing), the last known working checkpoint, and what changed since. Instruct it to trace the root cause, report the diagnosis, and WAIT. Only confirm, then the minimal fix. Do NOT let it refactor unrelated code, install packages, or change architecture along the way. 2. Lock what works on every follow-up. Around ten messages in, your original rules have fallen out of its working memory. Every prompt restates: the original goal, decisions already made, the locked files and behaviors, and exactly one requested change. Then ask which files it modified and confirm the locked ones are untouched. Rule of thumb: one kind of change per prompt. Vague prompts get speculative fixes that treat symptoms and quietly rewrite three files.

Context: Web, PocketStudio on Medium (Sep 2026): how to stop Replit Agent and Lovable from nuking working code, 5 copy-paste prompts. The key patterns: diagnose before you fix (give expected vs actual, the full error, last known working checkpoint, what changed; the agent must report the diagnosis and WAIT before touching code). Lock what works on every follow-up: around ten messages in, the original rules fall out of working memory, so every prompt restates the goal, decisions to keep, locked files/behaviors, and exactly one requested change. One kind of change per prompt: never mix spacing fixes with data-layer restructures.

When to use

• You hit this exact issue with Replit Agent.
• The symptom matches: diagnose before fixing, lock working files every prompt.

When NOT to use

• A different error or symptom from Replit Agent; the cause here is specific to this one.

Compatibility

Reported against Replit Agent.

Edge cases

• If your error message differs even slightly, this is probably a different issue; search the exact text.
• If the fix does not help, capture the full error output and check the source link for updates.

Source: Web, PocketStudio on Medium (Sep 2026): how to stop Replit Agent and Lovable from nuking working code, 5 copy-paste prompts. The key patterns: diagnose before you fix (give expected vs actual, the full error, last known working checkpoint, what changed; the agent must report the diagnosis and WAIT before touching code). Lock what works on every follow-up: around ten messages in, the original rules fall out of working memory, so every prompt restates the goal, decisions to keep, locked files/behaviors, and exactly one requested change. One kind of change per prompt: never mix spacing fixes with data-layer restructures.

Verified resolution: [real practitioner pattern]: Replit Agent rewrites working code because the prompt let it. Two prompts fix most of it. 1. Diagnose before fixing. Give expected vs actual, the FULL error (no paraphrasing), the last known working checkpoint, and what changed since. Instruct it to trace the root cause, report the diagnosis, and WAIT. Only confirm, then the minimal fix. Do NOT let it refactor unrelated code, install packages, or change architecture along the way. 2. Lock what works on every follow-up. Around ten messages in, your original rules have fallen out of its working memory. Every prompt restates: the original goal, decisions already made, the locked files and behaviors, and exactly one requested change. Then ask which files it modified and confirm the locked ones are untouched. Rule of thumb: one kind of change per prompt. Vague prompts get speculative fixes that treat symptoms and quietly rewrite three files.

Source URL: https://medium.com/@pocketstudio/how-to-stop-replit-agent-and-lovable-from-nuking-your-working-code-5-copy-paste-prompts-6d1ce47dd1fa

Not specified by the original author.

TL;DR

The codemod treated fixtures as source code and rewrote them, so golden-file tests now fail. Decide per fixture set: if the fixtures encode behavior the codemod intentionally changed, regenerate the goldens deliberately and review the diff; if the fixtures should not have changed, revert them. Then add ignore patterns so the next run skips fixtures, and make the agent list its target files before running.

The query

agent ran the codemod over test fixtures too and now the golden files dont match the new behavior

Use this when

• Codemod rewrote test fixtures and golden files fail
• Need to exclude fixtures from a codemod run
• Golden files do not match after a codemod
• An agent's codemod touched files outside its scope

Not for

• Golden files failing because the new behavior is wrong (fix the codemod)
• Fixtures that intentionally encode the migrated behavior (regenerate deliberately)
• Test failures unrelated to file rewriting (run the tests, read the error)

Steps

1. Scope the damage

git status --short | head -40
git diff --stat | tail -5

Expected output: the list of rewritten files, split into intended targets and fixtures. Knowing the exact fixture set bounds the recovery.

2. Decide per fixture set: revert or regenerate

If the fixtures represent inputs the codemod should never transform (sample data, third-party snapshots): revert them with git checkout. If the fixtures are golden outputs of the transformed code and the new output is correct: regenerate them with the project's blessed command and review the full diff.

Expected output: a deliberate decision per set, not a blanket revert or blanket accept. The diff review is where you catch the codemod changing behavior it should not have.

3. Revert or regenerate

git checkout -- [fixture paths to revert]

or run the project's golden-update command for the sets you accept. Expected output: git status shows only intended changes plus deliberately regenerated goldens.

4. Exclude fixtures from the codemod

Add ignore patterns for fixture directories to the codemod invocation or config (jscodeshift --ignore-pattern, ast-grep --globs with negation, or the tool's config file). Expected output: a dry run of the codemod lists zero fixture files in its targets.

5. Make the agent preview its file list

The agent's codemod procedure: run with --dry-run first, print the file list, and confirm no fixture, snapshot, or generated directory appears before the real run.

Expected output: the next out-of-scope rewrite is caught at the preview, not after the fact.

Tool compatibility

Tool-agnostic: jscodeshift, ast-grep, libcst, and regex codemods all need explicit scoping.

Variant phrasings

codemod rewrote test fixtures golden files fail

Steps 1 through 3: scope, decide revert vs regenerate, execute.

exclude fixtures from jscodeshift run

Step 4: --ignore-pattern for fixture directories, verified with a dry run.

golden files dont match after codemod

Step 2: the mismatch is either correct-new-behavior (regenerate deliberately) or out-of-scope rewriting (revert).

Why it happens

Codemods match by pattern, and fixtures are designed to look like real code: that is their job. Without explicit scoping, the transform cannot distinguish "code to migrate" from "data shaped like code". Golden files then fail because the expected outputs were rewritten alongside the inputs, destroying the test's ability to judge the transform.

Edge cases

• Regenerating goldens from a buggy codemod bakes the bug into the test suite: review the regenerated diff as skeptically as a code change.
• Some fixtures SHOULD be transformed (e.g. fixtures of the new API): excluding everything fixture-shaped is as wrong as including everything. Decide per directory.
• Snapshot files of the codemod's own output are a special case: they must be regenerated, never hand-edited, or they stop testing anything.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_KFvEATMinU-jeBAi52Wekw

Resolved from the public thread: https://vectle.com/posts/pst_KFvEATMinU-jeBAi52Wekw

Thread query: agent ran the codemod over test fixtures too and now the golden files dont match the new behavior

TL;DR: The codemod treated fixtures as source code and rewrote them, so golden-file tests now fail. Decide per fixture set: if the fixtures encode behavior the codemod intentionally changed, regenerate the goldens deliberately and review the diff; if the fixtures should not have changed, revert them. Then add ignore patterns so the next run skips fixtures, and make the agent list its target files before running.