Update Linear Post-Job

Update Linear Post-Job

When to Use

Use this skill when:

• Job completion (Tracking Agent): After QA validates implementation and Planning Agent confirms job complete
• Phase 7 of TDD workflow: Tracking Agent handles PR/documentation and Linear updates
• Work block tracking: Need to update child issue status and check parent completion

Triggers:

• Planning Agent confirms job completion
• QA Agent approves implementation
• PR is merged and job is ready for closure
• Need to track work block progress in Linear

TDD Workflow Context:

• 7-Phase TDD: Research → Spec → Linear Enrichment → QA (tests) → Action (code) → QA (validate) → Tracking (PR/docs) → Dashboard Update
• This skill: Executes during Phase 7 (Tracking Agent responsibilities)

Workflow

Step 1: Update Child Issue Status

Purpose: Mark the completed job (child issue) as Done in Linear

Prerequisites:

• Planning Agent confirmed job completion
• QA Agent approved implementation
• PR merged (if applicable)
• Handoff includes child issue ID

Execute:

// Update child issue to Done
await mcp__linear-server__update_issue({
  id: "LAW-5",  // Child issue ID from handoff
  state: "Done"
})

Then add completion comment:

await mcp__linear-server__create_comment({
  issueId: "LAW-5",
  body: `✅ **Job Complete**

**PR**: #14 (https://github.com/org/repo/pull/14)
**QA Status**: ✅ All tests passing
**Files Changed**: 8 files, +420/-15 lines

Ready for merge.`
})

Comment Template Fields:

• PR: Link to merged pull request (if applicable)
• QA Status: Confirmation that QA validation passed
• Files Changed: Summary of code changes
• Additional Notes: Any relevant completion details

Error Handling:

If update fails, add comment instead and continue:

try {
  await mcp__linear-server__update_issue({ id: "LAW-5", state: "Done" })
} catch (error) {
  await mcp__linear-server__create_comment({
    issueId: "LAW-5",
    body: `⚠️ **Tracking Agent**: Could not update issue status to Done.

Error: ${error.message}

Job completion confirmed by Planning Agent. Manual status update needed.`
  })
  // Continue to Step 2 - don't block on child update failure
}

Step 2: Check Parent Work Block Completion

Purpose: Determine if all child jobs in the work block are complete

Execute:

// 1. Get parent issue details
const parentIssue = await mcp__linear-server__get_issue({
  id: "LAW-4"  // Parent work block ID from handoff
})

// 2. Get all child issues of parent
const childIssues = await mcp__linear-server__list_issues({
  team: "Linear-First-Agentic-Workflow",  // From .project-context.md
  parentId: "LAW-4"  // Parent work block ID
})

// 3. Check if ALL children are Done
const allChildrenDone = childIssues.nodes.every(
  issue => issue.state.name === "Done"
)

Decision Point:

• If allChildrenDone === true: Proceed to Step 3 (update parent)
• If allChildrenDone === false: Skip to Step 4 (report to Planning Agent)

What to check:

• All child issues must have state.name === "Done"
• Check all siblings (other child issues under same parent)
• Do not update parent if any child is still In Progress, Todo, or Blocked

Step 3: Update Parent Work Block (If All Children Complete)

Purpose: Mark parent work block as Done when all child jobs complete

Execute only if Step 2 determined all children are Done:

// Update parent work block to Done
await mcp__linear-server__update_issue({
  id: "LAW-4",
  state: "Done"
})

// Add completion comment to parent
await mcp__linear-server__create_comment({
  issueId: "LAW-4",
  body: `✅ **Work Block Complete**

All child jobs completed:
- LAW-5: Research Agent Upgrade ✅
- LAW-6: Planning Agent Refactor ✅
- LAW-7: Tracking Agent Updates ✅
[... list all completed children ...]

All PRs merged. Work block ready for closure.`
})

Comment Template for Parent:

• List all completed child jobs
• Confirm all PRs merged
• Note that work block is ready for closure

Error Handling:

If parent update fails, report to Planning Agent but don't block:

try {
  await mcp__linear-server__update_issue({ id: "LAW-4", state: "Done" })
} catch (error) {
  // Report error to Planning Agent
  // Continue - job is still complete even if parent update fails
}

Step 4: Report to Planning Agent

Purpose: Notify Planning Agent of completion and any issues

Write completion report to handoff location (specified by Planning Agent):

If all updates successful:

**Tracking Agent Completion Report**

✅ **All Linear Updates Complete**

**Child Issue**: LAW-5 updated to Done
**Parent Work Block**: LAW-4 [updated to Done | still has N incomplete children]

**Summary**:
- PR #14 merged successfully
- Linear issues updated
- Work block [complete | in progress]

[If parent complete] Master Dashboard update delegated to Traycer per Phase 7 protocol.

Ready for next job.

If errors occurred:

**Tracking Agent Completion Report**

⚠️ **Linear Updates Completed With Errors**

**Child Issue**: LAW-5 - [updated successfully | update failed, manual intervention needed]
**Parent Work Block**: LAW-4 - [checked, N of M children complete | check failed]

**Errors**:
- [List any errors encountered]

**Action Required**:
- [List manual steps needed to resolve errors]

Job completion confirmed despite Linear update errors.

Important: Do NOT update Master Dashboard - this is delegated by Traycer per Phase 7 protocol

Reference

Complete Example Flow

Scenario: LAW-5 (Research Agent Upgrade) complete, parent is LAW-4

// Step 1: Update child issue
await mcp__linear-server__update_issue({
  id: "LAW-5",
  state: "Done"
})

await mcp__linear-server__create_comment({
  issueId: "LAW-5",
  body: "✅ Job complete. PR #14 merged."
})

// Step 2: Check parent completion
const parentIssue = await mcp__linear-server__get_issue({ id: "LAW-4" })

const childIssues = await mcp__linear-server__list_issues({
  team: "Linear-First-Agentic-Workflow",
  parentId: "LAW-4"
})

const allDone = childIssues.nodes.every(i => i.state.name === "Done")

// Step 3: Update parent if all children complete
if (allDone) {
  // All 9 child jobs complete
  await mcp__linear-server__update_issue({
    id: "LAW-4",
    state: "Done"
  })

  await mcp__linear-server__create_comment({
    issueId: "LAW-4",
    body: "✅ All 9 child jobs complete. Work block done."
  })
}

// Step 4: Report to Planning Agent
// Write completion report to handoff location
// Do NOT update Master Dashboard (delegated by Traycer)

Branch Creation Context

Note: This skill executes AFTER branch creation and PR merge

7-Phase TDD Workflow:

1. Phase 1: Research (no branch yet)
2. Phase 2: Spec clarification (no branch yet)
3. Phase 3: Linear Enrichment (no branch yet)
4. Just before Phase 4: Create branch
5. Phase 4-6: QA creates tests → Action implements → QA validates (on branch)
6. Phase 7: Tracking Agent handles PR/docs and Linear updates ← This skill

Branch Naming Convention (for reference):

Pattern: feat/<parent-issue-id>-<child-issue-id>-<slug>

Examples:

# Work Block: LAW-4, Child Job: LAW-5
feat/law-4-law-5-research-agent-upgrade

# Work Block: LAW-350, Child Job: LAW-351
feat/law-350-law-351-webhook-server-setup

Related Tools

• Linear MCP: mcp__linear-server__update_issue, mcp__linear-server__get_issue, mcp__linear-server__list_issues, mcp__linear-server__create_comment
• Project Context: .project-context.md (contains team name for Linear queries)
• Handoff Files: Planning Agent provides child/parent issue IDs

Related Documentation

• Original Reference: post-job-linear-updates.md (deprecated - use this skill instead)
• Agent Prompts:
  • Tracking Agent: docs/agents/tracking/tracking-agent.md
  • Planning Agent: docs/agents/planning/planning-agent.md
• Related Ref-Docs:
  • tdd-workflow-protocol.md - Full 7-phase TDD workflow
  • master-dashboard-creation-protocol.md - Dashboard update protocol (delegated by Traycer)
  • linear-update-protocol.md - General Linear API usage

Decision Matrix

Common Pitfalls to Avoid

❌ Don't: Update Master Dashboard directly

• Why: Dashboard updates are delegated by Traycer per Phase 7 protocol
• Instead: Report completion to Planning Agent, let Traycer coordinate dashboard update

❌ Don't: Block if parent update fails

• Why: Child job is complete regardless of parent status
• Instead: Report error to Planning Agent, mark job as complete

❌ Don't: Assume all children Done without querying

• Why: Another agent might have deferred a child job
• Instead: Always query all child issues and check state

❌ Don't: Update child issue state to "Closed" or "Canceled"

• Why: "Done" is the correct state for completed jobs
• Instead: Always use "Done" state

Version History

• v1.0 (2025-11-05): Converted from post-job-linear-updates.md to skill format
  • Extracted 3-step workflow from reference doc
  • Added error handling examples
  • Clarified decision points and delegation boundaries
  • Included complete example flow