/gaia-perf-deepdive

user-facing
Category:
Reviews
Lifecycle phase:
4 -- Implementation
Arguments:
[story-key]

What it does

Deep performance bottleneck analysis: N+1 query detection, memory and bundle impact assessment, caching gap analysis, and algorithmic complexity review. More thorough than /gaia-review-perf -- runs per-stack ORM pattern detection, complexity analysis, and static budget checks.

When to use it

  • You want a deep-dive performance analysis beyond the standard review gate.

Prerequisites

  • Story must be in review status (for story-based reviews). The story file must exist under .gaia/artifacts/implementation-artifacts/.

How to invoke

/gaia-perf-deepdive E3-S7

What it does step by step

  1. Resolve target Locates the story file or review target.
  2. Run deterministic analysis Executes per-stack tooling to collect evidence.
  3. Apply LLM judgment Performs semantic review on top of the deterministic evidence.
  4. Compute verdict The verdict (APPROVE, REQUEST_CHANGES, or BLOCKED) is computed by the verdict resolver -- the LLM does not determine the verdict.
  5. Update Review Gate Records the verdict in the story's Review Gate table.

Inputs

Story file, associated source files, and project configuration are read automatically.

Outputs

A review report is written to .gaia/artifacts/implementation-artifacts/. The story's Review Gate row is updated with PASSED or FAILED.

Example session

> /gaia-perf-deepdive E3-S7

Resolving story E3-S7...
Running deterministic analysis...
Applying semantic review...

Verdict: APPROVE
Review Gate updated: PASSED

0 critical, 1 warning, 2 suggestions.

What to run next

Troubleshooting

"Story must be in review status"

Run /gaia-dev-story to implement and transition the story first.

"Story file not found"

Verify the story key and check that the file exists under .gaia/artifacts/implementation-artifacts/.