Skills Library
Best AI skills using Linear
Three ready-to-run skills that turn Sentry errors and on-call noise into Linear issues your team can act on.
What are the best AI skills using Linear?
The best AI skills using Linear are Investigate Sentry Error, Track Sentry Error in Linear, and On-call Handoff Summary. Together they take an engineering team from a new error in Slack to a well-written Linear issue to a handoff the next on-call engineer can trust.
Linear is where engineering teams agree on what is real work. These three skills make sure the right things get there: errors are triaged before anyone files a ticket, tickets are only filed after a person approves, and the on-call handoff reads Linear for owners and status instead of guessing.
The type.com engineering team built them for its own on-call rotation. The public versions in the type.com Skills Library swap internal names for settings you fill in.
Browse them in the type.com Skills Library, or read how Linear works in type.com.
| Skill | Best for | What you get |
|---|---|---|
| Investigate Sentry Error | Engineers on call, the moment a Sentry error lands in a Slack channel | A short triage: impact, evidence, likely root cause, confidence, and one call: ignore, quick fix, or track |
| Track Sentry Error in Linear | Teams that want every real error in Linear, without duplicates or empty tickets | One Linear issue with problem, impact, evidence, proposed work, and acceptance criteria, or a link to the issue that already exists |
| On-call Handoff Summary | The outgoing on-call engineer, before every shift change, on a schedule | A draft handoff: active incidents, recurring problems, fixes in progress, what to watch, and what still needs a human |
Which Linear skills should an engineering team use?
The three skills follow an error from alert to handoff: investigate it, track it once a person approves, and hand it off at the end of the shift.
Investigate Sentry Error and Track Sentry Error in Linear are a pair: the first recommends, the second acts on a person's go-ahead. On-call Handoff Summary works on its own, and gets more useful once errors are tracked in Linear with clear owners.
1. Investigate Sentry Error
Investigate Sentry Error triages a Sentry error from its Slack thread: it reads the events, traces the stack into your code, checks Linear for existing work, and ends with one recommendation.
Most error threads end in a guess or a ticket nobody reads. This skill separates facts from hypotheses, writes Unknown instead of inventing impact or an owner, and searches Linear before suggesting new work. It never writes to Linear, so a noisy alert channel cannot flood your backlog.
- Best for
- Engineers on call, the moment a Sentry error lands in a Slack channel
- Works from
- The Slack thread, the Sentry issue and its events, your codebase, and a read-only Linear search
- You get
- A short triage: impact, evidence, likely root cause, confidence, and one call: ignore, quick fix, or track
2. Track Sentry Error in Linear
Track Sentry Error in Linear turns a human-approved Sentry error into one well-written Linear issue, or links the one that already exists, and backlinks the Sentry issue and Slack thread.
Auto-created tickets pile up as duplicates titled with raw exception text. This skill runs only after a person says "track this", searches Linear five different ways before writing, files new issues only in the team and project you configure, and reads the issue back to check it landed in the right place.
- Best for
- Teams that want every real error in Linear, without duplicates or empty tickets
- Works from
- A person's "track this" reply in the error thread, plus your Linear team and project for error work
- You get
- One Linear issue with problem, impact, evidence, proposed work, and acceptance criteria, or a link to the issue that already exists
3. On-call Handoff Summary
On-call Handoff Summary drafts the on-call handoff from a shift's Slack alerts, Sentry errors, Linear issues, and GitHub pull requests, with active incidents first and a source link on every item.
Handoffs written from memory miss the recurring bug and the fix that merged but never shipped. This skill groups every signal about one problem into a single item, takes owners and status from Linear, never calls a fix shipped just because it merged, and flags what it could not confirm.
- Best for
- The outgoing on-call engineer, before every shift change, on a schedule
- Works from
- Your alert channels in Slack, Sentry activity, Linear issues, and GitHub pull requests for the shift
- You get
- A draft handoff: active incidents, recurring problems, fixes in progress, what to watch, and what still needs a human
How do these skills keep Linear clean?
The skills keep a human between Sentry and Linear. Investigation is read-only, and tracking needs an explicit "track this" from a person.
Wiring Sentry straight into Linear fills the backlog with one ticket per alert. Here, Investigate Sentry Error ends with one of three calls. IGNORE comes with the evidence that should trigger a second look. QUICK FIX names who can act. TRACK ends with a line telling the team how to approve tracking.
When someone replies "track this", Track Sentry Error in Linear searches by Sentry ID, error message, stack frame, symptom, and likely root cause. If an issue already covers the same defect, it adds the missing Sentry and Slack links and says no duplicate was created. A bot's suggestion, or a question like "should we track this?", never counts as approval.
Example reply in the error thread
track this
- Investigation never creates, edits, or comments on a Linear issue.
- Tracking searches for duplicates before it writes anything.
- New issues go only to the team and project you configure.
- Every issue links back to the Sentry issue and the Slack thread.
How does the on-call handoff use Linear?
Run On-call Handoff Summary on a schedule before each shift change, and the outgoing engineer only has to review and add context.
The skill reads your alert channels, Sentry, Linear, and GitHub for the shift, then groups everything about one problem into a single item. Active incidents come first, each with impact, status, owner, next action, and links. Recurring problems, fixes in progress, and what to watch next shift follow.
It is careful about what it claims. A merged pull request is not called shipped without deployment evidence, owners come from Linear or are marked Unassigned, and anything uncertain goes under Needs human confirmation. The "How was on-call?" rating is left for the engineer to fill in.
- 1
Install and configure
Install On-call Handoff Summary from the library and set your alert channels and timezone.
- 2
Schedule it
Create an automation that runs the skill an hour before each shift change and posts to your on-call channel.
- 3
Review and hand off
The outgoing engineer corrects the draft, adds context, and fills in how the shift went.
How do engineering teams run Linear skills in type.com?
In type.com, an engineering team connects Linear, Slack, Sentry, and GitHub to a shared Space, installs the skills once, and runs them from Slack or on a schedule.
Install the skills from the library into your engineering Space, and everyone on the rotation gets the same triage format, the same Linear destination, and the same handoff. Mention @Type in an error thread to investigate, reply "track this" to file the issue, and let the handoff run on its own before each shift.
See how Linear connects to type.com, or browse every skill in the type.com Skills Library.
Frequently asked questions
What is an AI skill for Linear?
An AI skill is a packaged set of instructions, sometimes with scripts and reference files, that Claude or Codex follows the same way every time. A Linear skill is one that reads from or writes to Linear as part of that workflow, such as searching for existing issues, filing a new one, or pulling owners and status.
Will these skills create Linear issues on their own?
Only Track Sentry Error in Linear writes to Linear, and only after a person replies "track this" in the error thread. Investigate Sentry Error and On-call Handoff Summary only read Linear. None of them change status, priority, assignee, or labels on existing issues.
Can I choose which Linear team and project issues go to?
Yes. Before first use, you set the Linear team key and the exact project name for error work. The skill passes both explicitly when it creates an issue, and if either can't be found it stops and posts the draft issue in Slack instead of filing it somewhere else.
Which integrations do these skills need?
Linear and Slack for all three. Investigate Sentry Error and Track Sentry Error in Linear also need Sentry, and Investigate reads your codebase. On-call Handoff Summary reads Sentry and GitHub as well.
Do these skills work outside type.com?
Yes. The skills are plain SKILL.md folders, so they run in Claude Code, Codex, and other agents that read SKILL.md, as long as the agent can reach Linear, Slack, Sentry, and your repository.
Who made these Linear skills?
The type.com engineering team built them for its own on-call rotation and still uses them. The public versions replace internal team names, project names, and Slack channels with settings you fill in.