Allow pwd and date - #26
Merged
Merged
Conversation
Both were refused in a live run: the agent asked for `pwd` and `date` before doing anything else, got capability_denied for each, and spent a round trip rediscovering that `python3 -c "import os; print(os.getcwd())"` and `python3 -c "import datetime"` are the permitted spellings. At local model rates that is over a minute of wall clock each, plus the denial and the workaround in every subsequent prompt for the rest of the loop. Neither is a send vector, which is where the deny list draws its boundary, and both are strictly read-only. They belong with the other verification primitives. Also drops the hardcoded "32-entry" from _check_config_source's docstring, which this change would have made wrong and which the next one would have made wrong again; the count was never the point of the example.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What this changes
Both were refused in a live run: the agent asked for
pwdanddatebefore doing anything else, got capability_denied for each, and spent a round trip rediscovering thatpython3 -c "import os; print(os.getcwd())"andpython3 -c "import datetime"are the permitted spellings. At local model rates that is over a minute of wall clock each, plus the denial and the workaround in every subsequent prompt for the rest of the loop.Neither is a send vector, which is where the deny list draws its boundary, and both are strictly read-only. They belong with the other verification primitives.
Also drops the hardcoded "32-entry" from _check_config_source's docstring, which this change would have made wrong and which the next one would have made wrong again; the count was never the point of the example.
Checklist
make checkis green locallyare the product)
specs/(spec changes go through an issue first)Generated-by:trailer present if anagent wrote the change