agentleFS
ch4570ch4570/vulpora/skills/code-authoring-router/SKILL.md

code-authoring-router

Skill1 starsChanged 56 days agoApache-2.0View on GitHub

Route generic feature, bug-fix, refactor, and implementation requests to installed code-authoring skills after detecting repository stack evidence. Use even when the user only names the desired behavior and does not mention Kotlin, Spring, PostgreSQL, Microsoft SQL Server, or OpenSearch; do not use for review-only requests.

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 4 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Before editing, preserve the repository's current VCS state and inspect repository facts to select the construction guidance that matches …

    line 8
  2. Very specific:

    If root vulpora.config.json exists, validate and apply its locale and vcs policy.

    line 14
  3. Very specific:

    Do not create, switch, fetch, fast-forward, push, or open a change request merely because implementation was requested.

    line 18
  4. Very specific:

    Put base_branch, work_branch, required_authoring_skills, and evidence paths in plans or implementation handoffs so delegated agents apply …

    line 40
  5. Distinctive:

    Do not select a technology from the feature wording alone.

    line 42
  6. Distinctive:

    Read stack detection when evidence is mixed or the feature crosses multiple persistence/search boundaries.

    line 48
  7. Distinctive:

    Before completion, confirm that the configured VCS policy was preserved, every changed implementation lane used its available routed …

    line 52

≈ “Common” is an estimate: at least 1 in 5 of the 308,232 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Code authoring router

Before editing, preserve the repository's current VCS state and inspect repository facts to select the construction guidance that matches the affected code. A generic request such as “결제 기능 추가해줘” is an implementation signal, not proof of a technology stack.

Resolve project policy

If root vulpora.config.json exists, validate and apply its locale and vcs policy. Otherwise use the portable defaults: follow the user and repository language, auto-detect the hosted VCS and default branch only when needed, and preserve the current branch and worktree.

Do not create, switch, fetch, fast-forward, push, or open a change request merely because implementation was requested. Prepare a child branch only when the user/repository requires it or vcs.prepareBranch is true. In that opt-in path, use the configured provider and base branch (auto means the remote default, never an assumed develop), and preserve dirty or diverged work. The GitLab-specific git-flow skill may be used only when GitLab is proven and its house style matches the project policy.

Record the resolved policy and any branch action with the routing evidence.

Detect and route

  1. Inspect project instructions, build files, dependency catalogs, source extensions, migration directories, mappings, query builders, and nearby tests.
  2. Record the evidence paths and choose every installed, applicable authoring skill:
Repository and task evidenceLoad before editing
Actual Kotlin source in the affected module, supported by its build settings; the change edits Kotlin codekotlin-code-authoring
Java source in the affected module, including Java/SpringRepository-local Java authoring and test conventions. If the java-reviewer agent is installed, read its bundle's java-review/reference/kb/INDEX.md and only the matching topics as construction constraints, not as a review request; backend-test-author for Java tests and java-spring-review-workflow for Spring review if installed
PostgreSQL driver/dialect plus affected SQL, repository query, schema, or migrationpostgres-code-authoring
SQL Server driver/dialect plus affected T-SQL, stored procedure, schema, index, or migrationmssql-code-authoring
OpenSearch client/dependency plus affected mapping, Query DSL, pipeline, index setting, or search behavioropensearch-code-authoring
  1. Multiple skills may apply to one feature. Load each selected skill before the first edit in its lane and use only the relevant KB topics. If a matching skill is not installed, report the recommendation and continue with repository-local guidance rather than inventing a dependency.
  2. Put base_branch, work_branch, required_authoring_skills, and evidence paths in plans or implementation handoffs so delegated agents apply the same constraints.

Do not select a technology from the feature wording alone. If no supported stack has repository evidence, continue with repository-local conventions without inventing an authoring skill.

Gradle Kotlin DSL, buildSrc, and convention plugins do not establish Kotlin application usage. Exclude installed skill/agent bundles and evaluation fixtures from service-stack evidence. In mixed modules, route each affected file by its implementation/test language; Kotlin-only test skills must not be applied to Java work.

Read stack detection when evidence is mixed or the feature crosses multiple persistence/search boundaries.

Verify

Before completion, confirm that the configured VCS policy was preserved, every changed implementation lane used its available routed authoring skill, and the repository-provided formatter, lint, compile, and focused tests ran where applicable.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

Reviews can't be read right now.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.