Kotlin AI Agent Skills
A ready-made set of Agent Skills standard-compliant skill packages that teach coding agents the correct use of Kotlin, backend frameworks, and build tooling.
- Source repo
- Kotlin/kotlin-agent-skills
- Stars
- ★ 1.1k
- Last updated
- 1mo ago
- License
- Apache-2.0
- Primary language
- Shell
- FA score
- 42/100 · Major gaps
At a glance
- How it runs
- Works with
- Universal · cross-platformCodex · Claude Code
- Cost
- Free, no paid service needed
- Setup effort
- Low · running in minutes
- You'll need
- Typical use
- Backend developers using Kotlin with Spring/JPA who want agents to use these frameworks idiomatically — install a backend skill
- Not a fit if
- Teams needing stable, long-lived skill content (JetBrains incubator stage)
- Users whose coding agent lacks Agent Skills support
- Kotlin projects needing coverage beyond backend and build tooling
- Source review
- 42/100 · Major gaps 1 safety controls not found
What does this agent do, and when should you use it?
Kotlin/kotlin-agent-skills is a collection published by the official Kotlin GitHub organization for AI coding agents working on Kotlin projects. Each skill is a self-contained folder containing a SKILL.md file — YAML frontmatter with name and description followed by guidance the agent follows while the skill is active — conforming to the Agent Skills standard at agentskills.io. Skills follow the naming convention kotlin-<category>-<functional-name>, with current categories covering backend (correct use of Spring, JPA, and other server-side frameworks with Kotlin) and tooling (build tooling). Installation paths include the skills CLI (npx skills add), the Claude Code plugin marketplace, the Codex plugin marketplace, and manual copying into an agent's skills directory such as .claude/skills/. The project is Apache-2.0 licensed, carries an incubator badge, and proposes new categories via issues.
The repository ships installable skill folders: each skill contains a SKILL.md whose YAML frontmatter declares name and description, with body content giving instructions the coding agent follows when the skill is active. An example from the README is kotlin-tooling-agp9-migration, a tooling skill for migration scenarios. Users can install via npx skills add Kotlin/kotlin-agent-skills, via Claude Code commands (claude plugin marketplace add Kotlin/kotlin-agent-skills, then claude plugin install kotlin-agent-skills@Kotlin), via Codex (marketplace add in shell, then /plugins menu to search and install), or manually with cp -r skills/<skill> .claude/skills/. All skills live in the skills/ directory of the repository.
- Backend developers using Kotlin with Spring/JPA who want agents to use these frameworks idiomatically — install a backend skill
- Engineers migrating Android projects to AGP 9 — install kotlin-tooling-agp9-migration for migration guidance
- Teams on Claude Code who want the whole Kotlin skill set via the plugin marketplace in two commands
- Codex users adding the marketplace and installing the plugin through the /plugins menu
- Users with self-managed agent setups who copy individual skill folders from skills/ into .claude/skills/
- Kotlin community members contributing new skills or proposing new categories via issues
How do you install or deploy this agent?
Any one of:
1) skills CLI: npx skills add Kotlin/kotlin-agent-skills
2) Claude Code: claude plugin marketplace add Kotlin/kotlin-agent-skills, then claude plugin install kotlin-agent-skills@Kotlin
3) Codex: run codex plugin marketplace add Kotlin/kotlin-agent-skills in shell, then inside codex invoke /plugins, search kotlin-agent-skills, press Enter and select Install plugin
4) Manual: cp -r skills/kotlin-tooling-agp9-migration .claude/skills/ (copy the skill folder into your agent's skills directory)How do you use this agent?
Once installed, skills are discovered by the supported coding agent and activated on demand; when active, the agent reads the SKILL.md frontmatter (name, description) and body guidance and applies it to the task. You can also skip the marketplaces entirely and copy any folder from skills/ into your agent's skills directory (e.g. .claude/skills/).
What are this agent's strengths and limitations?
- Open Agent Skills standard (agentskills.io) compliance — not locked to one agent product; four install paths: skills CLI, Claude Code marketplace, Codex marketplace, manual copy
- Maintained by the official Kotlin organization with a clear naming convention kotlin-<category>-<functional-name> and defined categories (backend/tooling) that ease discovery and contribution
- Skills are self-contained folders (SKILL.md + resources), so you can pick and install a single skill, e.g. only the AGP9 migration skill
- Carries a JetBrains Incubator badge — early-stage project; skill content and structure may change
- Skills are Markdown guidance only; effectiveness depends on the host agent's Agent Skills support, with no standalone runtime
- Only two predefined categories (backend, tooling) — limited coverage, and new categories require opening an issue
- CLI install path requires Node.js/npx; Claude Code and Codex paths depend on each tool's plugin marketplace tooling
How does this agent compare with similar options?
The README names no competing Kotlin skill repositories, but because the skills follow the Agent Skills standard (agentskills.io), they are interoperable with any ecosystem supporting that standard; the differentiator here is official Kotlin-organization maintenance focused on Kotlin-specific scenarios.
Key facts side by side with the most closely related agents.
| Agent | Source review | Form / cost | Stars | Updated | Language | Full support on |
|---|---|---|---|---|---|---|
| Kotlin AI Agent Skills This agent | 42 · Major gaps | Agent plugin / skillFree | ★ 1.1k | 1mo ago | Shell | Codex · Claude Code |
| Pydantic AI Skills | 86 · Good | Library / SDKFree + model costs | ★ 380 | 7d ago | Python | OpenAI API |
| Finding-Unknowns Skills | 76 · Good | Agent plugin / skillFree | ★ 353 | 25d ago | Python | Claude Code · Claude.ai |
| EOC — Everything OpenAI Codex Workflow OS | 75 · Good | Agent plugin / skillFree + model costs | ★ 86 | 4mo ago | JavaScript | Codex |
How does FollowAgents rate this agent?
Why each dimension lost points
Evidence comprises only README, LICENSE, and a CI workflow; no SKILL.md or script content was provided, so what instructions the skills actually issue, least-privilege posture, confirmation needs, and sensitive-data handling cannot be verified — each gets 1 (inferred only from the 'instructions-only skill' nature, without direct support). rollback is 0: no uninstall/rollback guidance anywhere. source_attribution gets 2: the workflow credits its origin (Corey Haines) and the README carries a JetBrains incubator badge, but the LICENSE copyright line is an unfilled template placeholder, hence the deduction.
No skill content, tests, or validation output is available to assess consistency; the only visible reliability mechanism is validate-skills.yml (directory-name checks plus third-party validate-skill@v1), so all three criteria get 1: indirect, unexecuted corroboration only. The workflow depends on the external Action 'Flash-Brew-Digital/validate-skill@v1' and 'ubuntu-slim'; supply-chain and availability of that third-party Action cannot be verified statically.
Audience/scenario framing is reasonably clear (Kotlin projects, backend/tooling categories) and environment fit is covered by four install paths (skills CLI, Claude Code, Codex, manual copy), both scoring 2; capability_boundaries and trigger_precision rest entirely on SKILL.md frontmatter (name/description) not present in the evidence, so precision of trigger descriptions and stated boundaries cannot be confirmed — 1 each.
install_notes gets 3: README covers three official CLI installs plus manual copy with complete commands. license gets 3: full Apache-2.0 text in repo. information_architecture and naming_stability get 2: layout and naming convention are documented and CI-enforced, but the CATEGORIES file itself is absent from evidence. examples_and_faq and known_limitations get 1: no FAQ, no limitations section. versioning_changelog is 0: no version numbers, changelog, or release convention. maintenance_responsibility gets 1: only a pointer to CONTRIBUTING.md; maintainers and update path are not evidenced.
No skill body or output samples exist to judge output usability; marginal_value is a bare README assertion ('useful for Kotlin projects') without content support; cost_benefit cannot be estimated — all 1.
fact_inference_separation gets 2: README keeps factual claims (commands, layout) reasonably distinct from mechanism description without hype; claim_traceability gets 1: the 'follows the Agent Skills standard' claim has one external link and no local spec cross-check; cross_source_corroboration gets 1: correspondence with third-party CLI docs cannot be verified statically, and missing skill files make README-vs-content cross-validation impossible.
- Not found in source: rollback or recovery pathBack up first, or work on a git branch or snapshot, so its changes can be undone.
- This review saw no SKILL.md or script files; the actual skill instructions are unverified — read each skill's SKILL.md before installing and confirm it does not request excessive permissions or send data outward.
- CI depends on the third-party Action 'Flash-Brew-Digital/validate-skill@v1', a supply-chain risk; consider pinning it to a full commit SHA.
- The LICENSE copyright line is still the '[yyyy] [name of copyright owner]' placeholder and was never filled in.
- No versions or changelog exist, making it hard to pin or roll back installed skill content.
- The publisher is unverified in the enterprise registry and treated as unknown; do not infer safety from the Kotlin/JetBrains org name alone.