Battle-tested agent skills for real engineering workflows — git worktree discipline, end-to-end task implementation, PR review/fix loops, and the sharp edges of editing Jira/Confluence through an agent.
Every skill is a plain SKILL.md (+ optional scripts/ and agent-prompt templates), so they work with any skill-aware runtime: Claude Code, Kiro, or any agent that can read a markdown file and run shell commands. Paths inside each skill are written relative to the skill's own directory — no hardcoded home paths.
| Skill | Use when | Bundled |
|---|---|---|
| use-worktree | Creating git worktrees — sibling layout, safe teardown, stale-registration cleanup | create-worktree.sh |
| implement-task | Implementing a task end-to-end: Jira/plan/prompt → brainstorm → TDD in a worktree → review-fix loop → PR | 2 read-only agent templates |
| pr-review-inline | Reviewing a GitHub PR and posting findings as inline line comments (never a summary) with a per-finding selection gate | anchor-verifier contract |
| fix-pr-mantra | Addressing PR review comments — every comment gets a verdict, nothing is edited before the plan is approved | ledger-builder contract |
| jira-writing | Creating/editing Jira issues via the Atlassian MCP — ADF vs markdown traps, smartlink chips, EARS acceptance criteria | — |
| confluence-writing | Updating Confluence pages via the Atlassian MCP — whole-body replace hazards, WAF blocks, huge pages | — |
The six skills compose: implement-task drives use-worktree for isolation and a review loop before its PR; fix-pr-mantra closes the loop after review; pr-review-inline is the reviewer's side of the same table.
Dependencies:
pr-review-inlineandimplement-taskrequireuse-worktree(its bundled script does all worktree creation), andpr-review-inlinealso reads an agent template fromimplement-task. The plugin install brings all six; for manual installs, install those together.
/plugin marketplace add atomgunlk/skills
/plugin install atomgun
git clone --branch v1.0.0 --depth 1 https://github.com/atomgunlk/skills.git
cd skills
ln -s "$PWD/skills/engineering/<skill-name>" ~/.claude/skills/<skill-name>cp -R skills/engineering/<skill-name> ~/.kiro/skills/<skill-name>Copy the skill directory anywhere your agent can read, and reference it from your instructions file (AGENTS.md, rules, etc.):
When working with git worktrees, read and follow <path>/use-worktree/SKILL.md.Scripts resolve relative to the skill directory, so any location works.
Skills that touch company systems (pr-review-inline, and the Atlassian pair) start with a ## Personalize section — a small table of placeholders (your Jira site, ticket-key regex, gh wrapper, review voice) with the author's real setup shown as a worked example. Replace the example column with your values before first use; everything else is generic.
Each skill came out of real production use, then a verification pass: shell scripts are live-tested (contract: result on stdout, diagnostics on stderr), inter-file contracts are cross-checked, and MCP parameter names are validated against live tool schemas. The authoring approach follows the TDD-for-documentation methodology from obra/superpowers (writing-skills).