For AI agentsA Markdown index of these docs is at /llms.txt. Every page is also Markdown: add .md to its address.

Docs / The five questions

check_my_changes

After editing: callers you didn't update, and anything we can't see.

Your agent asks it after editing.

What it takes

InputTypeWhat it is
pathslist of text, optional—
teststrue or false, optionaltests: true also lists the tests that call the changed code (off by default).
expandtext, optionalA w-handle from this session to list in full, or reset.
formattext or json, optional—
familyyes, no or unsure, optionalfamily: your answer to 'are all subclasses of these types inside this repository?' yes, no or unsure (default unsure). It sets the [a] assumption in answers.

Example

The example project's Store.save gained a required parameter, and the file was saved.

your agentcheck_my_changes(paths: ["app/store.py"])
llmgraph · check_my_changes: 1 file changed, 2 calls break
[delta] files changed since your last call: app/store.py: Store body differs; Store.save signature differs (parameters, types)
⚠ breaks if it runs Store.save app/admin.py:3 reset: misses "flag" (within_these, one of 1, closed: repo only) [c] w17
⚠ breaks if it runs Store.save app/service.py:4 persist: misses "flag" (within_these, one of 3, closed: repo only) [a] w18
⚠ may break app/service.py:8 persist_any (open: type unknown) w19
⚠ plus dynamic calls elsewhere that could reach Store.save w20
assumes [a]: all subclasses are in this repo (project setting)
assumes [c]: our own reading, not confirmed by the checker
captured from the engine on the example project

Each line explained: Reading an answer.

Optionally, the answer can end with one group, "tests that call the changed code": one line per test file with its number of calls and its weakest label (up to four files, then a count line with a handle), so your agent can re-run those tests first. It is off by default. Your agent asks for it with tests: true, or you switch it on for every call with the setting tests = "on" under [check] in your settings file; any other value means off.

How fast

Once LLMGRAPH has caught up with a save (a second or two on the projects we tested), check_my_changes answers in well under a second.

What it cannot tell you

  • It checks argument count and names, not types, for Python.
  • TypeScript type clashes read "may break" until the type checker confirms.
  • It skips *args and spread arguments, and decorated functions.
  • It sees only saved files.
For agentsthis page as Markdown/docs/questions/check_my_changes.mdllms.txt