A coding agent can inspect a repository. It cannot reliably infer which behavior the product owner intended. Before handoff, distinguish implementation choices from product decisions.
“Reuse the existing project permission checks” is useful context. “Make it secure” leaves the access rules unspecified. “Use the existing creation flow” narrows scope; an invented file path sends the agent in the wrong direction.
Include the expected behavior, relevant entry point if known, explicit exclusions, dependencies and testable acceptance criteria. Ask the agent to surface unresolved decisions before changing code. Keep credentials and production customer data out of issue examples.
For a UI change, specify permissions, loading and error states, keyboard interaction and verification. For a bug, include reproduction steps. Start with the reusable templates and adapt them to your team's actual behavior.
SpecBot supplies a readiness review inside Linear. Use its findings to improve the issue, then hand it to your engineer or coding agent. SpecBot does not block an external agent from running and its verdict is not a test result.
Actual local review output
Synthetic example · actual offline fallback review. This is not a hosted AI result, customer outcome or live Linear screenshot. Scores vary by engine and input.
Input: Fix login bug — Users say login is broken. Need a fix soon. Label: bug; priority: unset.
🔴 NOT READY FOR AI IMPLEMENTATION — 23/100
Status: NOT_READY · Blockers: 4 · Improvements: 3 · Threshold: 80
This issue is missing context that could cause a coding agent to guess or produce rework.
Blockers
- State the expected behavior or outcome after the work is complete.
- Add explicit acceptance criteria that define done.
- Add reliable reproduction steps with actual vs expected behavior.
- Set priority and clarify the scope boundaries for this issue.
Suggested fixes
- Add technical context such as affected systems, APIs, files, or constraints.
- Document edge cases, failure modes, or exceptions.
- List dependencies, blockers, or related work.
Recommended next action
Add the missing information below and re-run SpecBot before handing this issue to a coding agent.
After fixes: Run SpecBot again before delegating: mention @SpecBot and say recheck.
Context to confirm
- [ ] Describe the expected behavior clearly
- [ ] Add acceptance criteria that define done
- [ ] Cover key edge cases and dependencies
- Add the affected systems, APIs, files, or services.
- Note any dependencies or blockers.
- Include reproduction details if this is a bug.
Score breakdown
- Objective / problem: 34/100
- Expected behavior: 14/100
- Acceptance criteria: 12/100
- Edge cases: 10/100
- Technical context: 44/100
- Dependencies: 16/100
- Reproduction steps: 8/100
- Priority and scope clarity: 42/100
*Confidence: 55% · Settings · Powered by SpecBot*
Install SpecBot in Linear and check an issue before delegation.