chore: Create a skill file that captures the wizard architecture principles#438
Merged
Conversation
🧙 Wizard CIRun the Wizard CI and test your changes against wizard-workbench example apps by replying with a GitHub comment using one of the following commands: Test all apps:
Test all apps in a directory:
Test an individual app:
Show more apps
Results will be posted here when complete. |
gewenyu99
approved these changes
May 6, 2026
Collaborator
gewenyu99
left a comment
gewenyu99
left a comment
There was a problem hiding this comment.
Some things to think about:
- Somewhere, maybe many places, we should tell it to update this if it makes conflicting hcnages
- What about tests? What needs tests?
- We should run this on PostHog Code with cloud runner on 4.7 max and let it do some iterative refactoring. I know for a fact we break these rules some places (probably my fault too)
|
|
||
| **New agent tool (in-process MCP):** Add it to `wizard-tools.ts` alongside `check_env_keys`, `set_env_values`, etc. The tool runs locally — secret values never leave the machine. Register the tool name in `WIZARD_TOOL_NAMES` so the SDK allowlist includes it. Follow the existing tool pattern: zod schema, path-traversal protection, logging. | ||
|
|
||
| **New security rule:** Add a `YaraRule` object to the `RULES` array in `yara-scanner.ts`. Each rule has a name, description, severity, category, `appliesTo` (which hook+tool combinations), and `patterns` (compiled regex). One match per rule is sufficient. The hooks in `yara-hooks.ts` don't need to change — they iterate the rules array automatically. |
Collaborator
There was a problem hiding this comment.
Is this the right extension surface always or should some go in the warlock repo?
|
|
||
| These are the early warning signs that a change is drifting from the discipline: | ||
|
|
||
| - **The runner is getting longer.** If you're adding lines to `agent-runner.ts` or `agent-interface.ts`, ask whether the concern belongs in a `WorkflowRun` config, a middleware, a post-run hook, or a skill file. |
Collaborator
There was a problem hiding this comment.
Almost nothing should go into the agent runners anymore. I find that this is a common trip up for agents
Co-authored-by: Vincent (Wen Yu) Ge <29069505+gewenyu99@users.noreply.github.com>
Collaborator
Author
|
@gewenyu99 I just added a whole skill maintenance reference, plus guidance in the |
daniloc
changed the title
Collaborator
Author
|
plus a note about tests in |
sarahxsanders
approved these changes
May 11, 2026
Collaborator
sarahxsanders
left a comment
sarahxsanders
left a comment
There was a problem hiding this comment.
warlock stuff is gooooooood 🚀
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
3 participants
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.
Let's make it easy for other teams to extend the wizard, with pull requests that are easy to approve.
This skill is a collaboration between myself and Opus 4.7:
After extended discussion, the model took a swing at
SKILL.md, which I edited closely. I think it's pretty good!ARCHITECTURE.mdcues future agents with a low-cost survey of the project, sparing it around a dozen tool calls of reading.ANTI-PATTERNS.mdilluminates the failure modes of breaking the architecture bargains.Questions
ANTI-PATTERNS? It certainly follows the redirect strategy we know works well at a micro level (instead ofuseEffect, do…)