Skip to content

test(prompts): add PM terminology, duplicate, VerifyChanges, and debug gadget tests#991

Merged
zbigniewsobiecki merged 1 commit intodevfrom
feature/prompt-rendering-consistency-tests
Mar 23, 2026
Merged

test(prompts): add PM terminology, duplicate, VerifyChanges, and debug gadget tests#991
zbigniewsobiecki merged 1 commit intodevfrom
feature/prompt-rendering-consistency-tests

Conversation

@aaight
Copy link
Copy Markdown
Collaborator

@aaight aaight commented Mar 23, 2026

Summary

  • Adds PM terminology rendering tests for splitting and planning agents (JIRA issue vs Trello card, JIRA URL examples, JIRA subtask description note)
  • Adds default (no pmType) Trello terminology fallback tests
  • Adds duplicate content detection tests for all 10 agent types — excludes content inside fenced code blocks (code examples legitimately repeat patterns)
  • Adds VerifyChanges presence tests for respond-to-ci and respond-to-review agents
  • Adds debug agent gadget naming tests verifying ListDirectory, RipGrep, and Tmux are capitalized correctly

Test plan

  • PM terminology: splitting with pmType=jira renders "issue" and atlassian.net URL
  • PM terminology: planning with pmType=jira renders "JIRA", "issue", JIRA subtask note, and atlassian URL
  • PM terminology: default (no pmType) falls back to Trello (card, trello.com/c, "Trello")
  • Duplicate content: all 10 agent types produce no duplicate block of 3+ consecutive non-trivial, non-code-block lines
  • VerifyChanges: respond-to-ci and respond-to-review prompts contain "VerifyChanges"
  • Debug gadgets: debug prompt contains ListDirectory, RipGrep, Tmux
  • All 6323 existing tests still pass
  • Lint and type checks pass

Trello card: https://trello.com/c/69c12720d72b1358bbe396bd

🕵️ claude-code · claude-sonnet-4-6 · run details

Copy link
Copy Markdown
Collaborator

@nhopeatall nhopeatall left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM — Clean, well-structured test additions that guard prompt rendering correctness across four distinct concerns (PM terminology, duplicate content, VerifyChanges presence, and gadget naming). The findDuplicateBlocks helper is thoughtfully designed with good exclusions for code blocks and trivial lines. All CI checks pass.

🕵️ claude-code · claude-opus-4-6 · run details

@zbigniewsobiecki zbigniewsobiecki merged commit 5faa421 into dev Mar 23, 2026
9 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants