feat: Introduce company-wide prompt hub - #1
Closed
manana2520 wants to merge 1 commit into
Closed
Conversation
- Add detailed README.md explaining the repository's structure, contribution guidelines, and how to use the prompts. - Create a 'coding-assistants' directory with a guide on syncing Claude agents locally using two different methods (sync script and sparse-checkout). - This initial commit establishes the foundation for a centralized and collaborative prompt library.
eddycek
added a commit
that referenced
this pull request
Oct 11, 2025
Copilot Review Issue #1 & #2: - Removed 'Bash(git reset:*)' from allow list to prevent override of 'git reset --hard:*' in ask list - Removed 'Bash(git push:*)' from allow list to prevent override of 'git push --force:*' and 'git push -f:*' in ask list Now: - git reset --hard requires confirmation (was being auto-approved) - git push --force/-f requires confirmation (was being auto-approved) - Regular git push and git reset are no longer auto-approved (will ask) This ensures destructive operations always require explicit confirmation. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
eddycek
added a commit
that referenced
this pull request
Oct 11, 2025
Copilot Review Issue #1 & #2: - Removed 'Bash(git reset:*)' from allow list to prevent override of 'git reset --hard:*' in ask list - Removed 'Bash(git push:*)' from allow list to prevent override of 'git push --force:*' and 'git push -f:*' in ask list Now: - git reset --hard requires confirmation (was being auto-approved) - git push --force/-f requires confirmation (was being auto-approved) - Regular git push and git reset are no longer auto-approved (will ask) This ensures destructive operations always require explicit confirmation. 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
4 tasks
jordanrburger
added a commit
that referenced
this pull request
Jul 19, 2026
Mined first user messages from real Kai conversations (OTEL trace analysis tables in the Keboola AI project, fct_otel_conversations) and turned the recurring intent patterns into activation cases. Every utterance is an anonymized paraphrase — client names, project IDs, URLs and client-specific table names replaced with generic equivalents; nothing is copied verbatim. What real traffic contributed: - job-error debugging is the dominant real intent (the 'Resolve this job error' deep-link) — new debug-component positives for output- mapping mismatches, OAuth grant failures, python tracebacks, plus the SQL-transformation-error hard negative (the #1 real-world boundary: transformation errors are not Python-component debugging) - transformation/config editing, flow scheduling, conditional flows -> keboola-config; security-finding fixes, broken-flow audits, SQL review -> keboola-cli (with the transformation-review overlap between the two captured as deliberate boundary cases) - data-app creation/preview/perf asks -> dataapp-development; a real 'use the keboola managed repo, I have no github account' -> keboola-git, with 'set up GitHub integration for our flows' as its hard negative - deprecated-component config migration as a hard negative for BOTH migrate-to-uv and develop-component/get-started (config migration is not code migration) - ~a quarter of real traffic is Czech/Slovak — added cs/sk utterances across skills to test routing across languages Two gold corrections surfaced by the live run: converting an existing script / building a brand-new custom component is initial scaffolding (get-started) per develop-component's own boundary, so those are negatives for develop-component. Live run: 239 cases, 94.1% overall accuracy, VERDICT: PASS. The 14 misroutes are the sharpened description-tuning backlog — notably the keboola-cli vs keboola-config transformation-review overlap and 'conditional flow' attracting build-component-ui. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
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
1 participant
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.
This pull request introduces the initial structure and documentation for the company-wide prompt hub.