@willbooster/agent-skills 1.28.4 → 1.28.6

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@willbooster/agent-skills",
-  "version": "1.28.4",
+  "version": "1.28.6",
   "description": "A collection of agent skills for WillBooster Inc.",
   "repository": {
     "type": "git",
@@ -48,7 +48,7 @@
     "@typescript/native-preview": "7.0.0-dev.20260518.1",
     "@willbooster/oxfmt-config": "1.2.2",
     "@willbooster/oxlint-config": "1.4.6",
-    "@willbooster/wb": "13.19.3",
+    "@willbooster/wb": "13.19.7",
     "build-ts": "17.1.15",
     "conventional-changelog-conventionalcommits": "9.3.1",
     "lefthook": "2.1.6",

package/skills/complete-pr/references/other-repositories-workflow.md

@@ -1,5 +1,6 @@
 1. Check the CI (GitHub Actions) results using the `check-pr-ci` skill.
-2. If any workflow fails, pull the latest remote changes, resolve the issues, commit, push, and return to step 1. Otherwise, proceed to the next step.
+2. If any workflow fails, pull the latest remote changes, investigate the root cause of the issues, resolve the issues, commit, push, and return to step 1. Otherwise, proceed to the next step.
+   - If this is a `wbfy` application branch, investigate the root cause to determine whether the failure is related to `wbfy`. If needed, fix `wbfy` itself as described in the `wbfy` skill.
 3. Fetch unresolved review threads using the `manage-pr-review-threads` skill.
 4. Review each unresolved thread and decide whether it requires a code or documentation change. Treat every review claim as unverified until you confirm it with concrete evidence from the current codebase, a local reproduction, or official documentation. Do not make changes just because a reviewer asserted them; make changes only after the evidence shows the comment is valid. Ignore only comments that are clearly outdated, incorrect, out of scope, or intentionally declined with solid reasoning.
 5. If there are valid review comments to address, make only the changes supported by the evidence from step 4, commit, push, and then return to step 1.

package/skills/complete-pr/references/willbooster-workflow.md

@@ -1,5 +1,6 @@
 1. Check the CI (GitHub Actions) results using the `check-pr-ci` skill.
-2. If any workflow fails, pull the latest remote changes, resolve the issues, commit, push, and return to step 1. Otherwise, proceed to the next step.
+2. If any workflow fails, pull the latest remote changes, investigate the root cause of the issues, resolve the issues, commit, push, and return to step 1. Otherwise, proceed to the next step.
+   - If this is a `wbfy` application branch, investigate the root cause to determine whether the failure is related to `wbfy`. If needed, fix `wbfy` itself as described in the `wbfy` skill.
 3. Fetch unresolved review threads using the `manage-pr-review-threads` skill.
 4. Review each unresolved thread to determine if it requires code or documentation changes. Treat every review claim as unverified until you confirm it with concrete evidence from the current codebase, a local reproduction, or official documentation. Do not make changes just because a reviewer asserted them; make changes only after the evidence shows the comment is valid.
 5. Address valid review comments by updating the code. For invalid comments, add explanatory comments to the code detailing why the existing implementation is necessary.

package/skills/review-fix-all/SKILL.md

@@ -8,7 +8,7 @@ allowed-tools: Bash(bunx:*), Task
 
 1. Run the following command with a 1-hour timeout (RUN IT AS A BLOCKING, FOREGROUND TASK TO PREVENT PREMATURE TERMINATION): `bunx @willbooster/agent-skills@latest review --agent all`
 2. Merge the returned findings, deduplicating those that address the same underlying issue. If there are no findings, quit without modifying the code.
-3. Accumulate every returned finding as a fact inside `## Verified Facts`, together with any returned `## Verified Facts` sections. Keep only facts that still affect review: what is known, why it matters, and how it was verified.
+3. Accumulate only review-meaningful facts inside `## Verified Facts`, together with any useful returned `## Verified Facts` sections. Keep facts that help avoid duplicate review work, such as current-code evidence, accepted constraints, or primary-source behavior; omit routine verification results and review-process history.
 4. Refine accumulated facts to reduce duplication and remove stale details, then pass them through stdin with heredoc syntax:
 
    ```bash

package/skills/review-fix-claude/SKILL.md

@@ -8,7 +8,7 @@ allowed-tools: Bash(bunx:*)
 
 1. Run the following command with a 1-hour timeout (RUN IT AS A BLOCKING, FOREGROUND TASK TO PREVENT PREMATURE TERMINATION): `bunx @willbooster/agent-skills@latest review --agent claude`
 2. If the command returns `There is no concern.`, quit without modifying code. Otherwise, process the returned findings.
-3. Accumulate every returned finding as a fact inside `## Verified Facts`, together with any returned `## Verified Facts`. Keep only facts that still affect review: what is known, why it matters, and how it was verified.
+3. Accumulate only review-meaningful facts inside `## Verified Facts`, together with any useful returned `## Verified Facts`. Keep facts that help avoid duplicate review work, such as current-code evidence, accepted constraints, or primary-source behavior; omit routine verification results and review-process history.
 4. Refine accumulated facts to reduce duplication and remove stale details, then pass them through stdin with heredoc syntax:
 
    ```bash

package/skills/review-fix-codex/SKILL.md

@@ -8,7 +8,7 @@ allowed-tools: Bash(bunx:*)
 
 1. Run the following command with a 1-hour timeout (RUN IT AS A BLOCKING, FOREGROUND TASK TO PREVENT PREMATURE TERMINATION): `bunx @willbooster/agent-skills@latest review --agent codex`
 2. If the command returns `There is no concern.`, quit without modifying code. Otherwise, process the returned findings.
-3. Accumulate every returned finding as a fact inside `## Verified Facts`, together with any returned `## Verified Facts`. Keep only facts that still affect review: what is known, why it matters, and how it was verified.
+3. Accumulate only review-meaningful facts inside `## Verified Facts`, together with any useful returned `## Verified Facts`. Keep facts that help avoid duplicate review work, such as current-code evidence, accepted constraints, or primary-source behavior; omit routine verification results and review-process history.
 4. Refine accumulated facts to reduce duplication and remove stale details, then pass them through stdin with heredoc syntax:
 
    ```bash

package/skills/review-fix-gemini/SKILL.md

@@ -8,7 +8,7 @@ allowed-tools: Bash(bunx:*)
 
 1. Run the following command with a 1-hour timeout (RUN IT AS A BLOCKING, FOREGROUND TASK TO PREVENT PREMATURE TERMINATION): `bunx @willbooster/agent-skills@latest review --agent gemini`
 2. If the command returns `There is no concern.`, quit without modifying code. Otherwise, process the returned findings.
-3. Accumulate every returned finding as a fact inside `## Verified Facts`, together with any returned `## Verified Facts`. Keep only facts that still affect review: what is known, why it matters, and how it was verified.
+3. Accumulate only review-meaningful facts inside `## Verified Facts`, together with any useful returned `## Verified Facts`. Keep facts that help avoid duplicate review work, such as current-code evidence, accepted constraints, or primary-source behavior; omit routine verification results and review-process history.
 4. Refine accumulated facts to reduce duplication and remove stale details, then pass them through stdin with heredoc syntax:
 
    ```bash

package/skills/wbfy/SKILL.md

@@ -8,7 +8,7 @@ allowed-tools: Bash(bun:*), Bash(gh:*), Bash(git:*), Bash(yarn:*)
 
 1. If the current branch is `main`, create and switch to a new branch based on the latest remote `main` branch.
    - Ensure the new branch is derived from the latest remote `main` branch to prevent merge conflicts when opening a pull request.
-2. Make sure .env files define NEXT_PUBLIC_BASE_URL correctly, as `wbfy` uses this environment variable to generate configuration files.
+2. Add or fix `NEXT_PUBLIC_BASE_URL` only if app code, tests, or existing Playwright config require a public app origin; Playwright alone is not enough.
 3. Update `~/ghq/github.com/WillBooster/shared/packages/wbfy` (a git repository) to the latest `main` branch.
 4. Run `yarn start <root_path_of_target_repo>` in `~/ghq/github.com/WillBooster/shared/packages/wbfy`.
 5. Run `yarn verify` or `bun verify` in the target repository (not within the `wbfy` directory).