Skills
Your playbooks, runnable on any pull request.
A security sweep, your team's review checklist, a changelog writer. Talyn finds the skills you already have — in the repository, in ~/.claude/skills, or saved to your workspace — and runs one against any pull request with a click.
- Picks up SKILL.md files from the repository, your machine, and your workspace
- Searchable picker with your most-used skills on top
- Output lands on the pull request: one review comment, or commits to the branch
- The same format Claude Code uses — if you have skills, they already work
Free plan: skills themselves are uncapped. Running one is an ordinary cloud task, so the three-concurrent-task limit is the brake.
feat: streaming results in the dashboard
sundial/web#412 · @dana · opened 2h ago
fix: retry the flaky checkout webhook
sundial/api#418 · @theo · opened 5h ago
perf: cache the search index
sundial/web#407 · @dana · opened 6h ago
Run a skill on sundial/web#412
Frequently used
pr-reviewRepo×12
Deep review: correctness, tests, and API surface
security-sweepTalyn×7
Scan the diff for authz gaps and injection risks
changelog-entryLocal
Draft the changelog entry from the diff
perf-auditRepo
Flag N+1s and hot-path allocations in the change
You have written the instructions already
Most teams have a review checklist. Somebody has written down what to look for in a migration, or what a good changelog entry contains, or the six things that have caused an incident before. It lives in a wiki page nobody opens, or in one person's head.
A skill is that written down in the standard SKILL.md format — the one Claude Code uses — and made runnable. Pick a pull request, pick a skill, and a cloud agent follows the playbook against that change.
feat: streaming results in the dashboard
sundial/web#412 · @dana · opened 2h ago
fix: retry the flaky checkout webhook
sundial/api#418 · @theo · opened 5h ago
perf: cache the search index
sundial/web#407 · @dana · opened 6h ago
Run a skill on sundial/web#412
Frequently used
pr-reviewRepo×12
Deep review: correctness, tests, and API surface
security-sweepTalyn×7
Scan the diff for authz gaps and injection risks
changelog-entryLocal
Draft the changelog entry from the diff
perf-auditRepo
Flag N+1s and hot-path allocations in the change
Three places, no setup
Talyn looks for skills where they already are rather than asking you to import them. Committed in the repository, so the whole team gets them and they version with the code. Sitting in `~/.claude/skills` on your machine, which is where yours probably already are. Or saved to your workspace, for the ones that are not repository-specific and are not just yours.
One caveat worth knowing before you rely on it: reading `~/.claude/skills` needs the desktop app, because the browser cannot see your disk. Skills used by a workflow or a loop have to be repository or workspace ones for the same reason — those run on the server, with the app closed.
- In the repository — versioned with the code, shared with the team
- In ~/.claude/skills — desktop app only
- Saved to the workspace — shared, and reachable by workflows and loops
What comes back
It depends what the skill asks for. A review-shaped skill posts a single review comment on the pull request. A fix-shaped skill commits to the branch. Either way it is one artefact, not a running commentary — the same restraint the code review feature is built around.
Skills compose with the rest of Talyn: a workflow can run one automatically when a pull request opens, and a loop can run one on a schedule.
Questions, answered.
Saved instructions you can re-run — a review checklist, a security pass, a changelog writer. Write one once and run it on any pull request with a click. Talyn finds the ones already in your project or on your machine, in the standard SKILL.md format, so if you use Claude Code you likely have some already.
No, it is the standard SKILL.md format. A skill is a markdown file with some frontmatter. Nothing about it is Talyn-specific, and skills you write for Talyn work in Claude Code too.
Yes, as long as it is a repository or workspace skill. A skill that only exists in ~/.claude/skills is refused when you save the rule, because the backend genuinely cannot read your machine — a rule that failed silently at 3am would be worse than one that refuses while you are looking at it.
Talyn does this for you.
Mission control for your GitHub pull requests, running on the Claude or ChatGPT subscription you already pay for. Free for three tasks at a time.
Keep reading
Feature
Code review
Findings in the app, not on your PR
Read moreFeature
Workflows
Rules that run themselves
Read moreFeature
Fix a PR with an agent
Failing tests, conflicts, review comments
Read moreGuide
Multiple Claude Code PRs
Running several coding agents at once produces pull requests faster than you can land them. Why they go stale in a predictable order, why updating each branch makes it worse, and what to do instead.
Read the guide