Fix argument interpolation and add behavioral evals #4
Conversation
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
Command prompts referenced arguments as vague prose ("if auto-fix is
enabled") instead of using $ARGUMENTS.<name> at decision points. The
executing agent could miss interpolated values and fall back to safe
defaults like auto-fix=false.
Replace all prose references with explicit $ARGUMENTS.* tokens in the
instruction body. Add a static test enforcing every frontmatter argument
appears as $ARGUMENTS.<name> in the body, and promptfoo behavioral evals
that verify models correctly interpret argument values.
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Outdated suggestions cannot be applied.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.
Summary
$ARGUMENTS.<name>at decision points. The executing agent could miss interpolated values and fall back to safe defaults (e.g.,auto-fix=false). All prose references replaced with explicit$ARGUMENTS.*tokens in the instruction body.make eval) that send interpolated prompts to Claude and assert the model correctly interprets argument values (auto-fix true/false, passes count, model resolution, scope resolution, tidy flags).test_argument_interpolationintest.shenforces every frontmatter argument appears as$ARGUMENTS.<name>in the body — prevents future regressions.Test plan
make test— 51 structural tests pass (including new interpolation check)make eval— 8 behavioral evals pass withANTHROPIC_API_KEYset$ARGUMENTS.*reference from a command body causesmake testto fail