explain-plan-analyzer

name: explain-plan-analyzer description: Analyzes PostgreSQL (and YugabyteDB) EXPLAIN / EXPLAIN ANALYZE query plans to identify performance issues. Use this skill whenever a user pastes an explain plan, asks to analyze a query plan, mentions seq scans or full table scans, wants index recommendations, or asks why a query is slow. Trigger even if the user just says "look at this plan", "what's wrong with this query", or pastes raw EXPLAIN output.

Explain Plan Analyzer

When a user provides a PostgreSQL or YugabyteDB EXPLAIN or EXPLAIN ANALYZE plan, work through these stages in order: parse → identify issues → recommend fixes → present clearly.

Stage 1: Understand the Plan Type

First determine what you're working with:

• EXPLAIN only — has cost estimates but no actual timing or row counts
• EXPLAIN ANALYZE — has both estimates and actuals (actual time=, actual rows=, loops=)
• EXPLAIN (ANALYZE, BUFFERS) — also includes I/O hit/miss data

Note this at the start of your response, as it affects how confident you can be about findings.

Stage 2: Parse Key Node Attributes

For each node in the plan tree, extract where present:

• Node type: Seq Scan, Index Scan, Index Only Scan, Bitmap Heap Scan, Hash Join, Nested Loop, Merge Join, Sort, Hash, Aggregate, etc.
• Relation name: the table or index being accessed
• rows: planner's estimated row count
• actual rows: real rows returned (ANALYZE only)
• cost: startup..total cost
• actual time: startup..total ms per loop (ANALYZE only)
• loops: how many times this node executed (ANALYZE only)
• filter / index cond / recheck cond: the predicate applied
• rows removed by filter: how many rows were scanned but discarded
• Buffers hit/read: I/O activity (BUFFERS only)

Stage 3: Seq Scan Analysis (Highest Priority)

Seq Scans are the first and most critical thing to evaluate.

Significance Thresholds

Flag a Seq Scan as significant (🔴 Critical) if ANY of these apply:

Flag a Seq Scan as minor (🟡 Low priority) if:

• Estimated rows < 1,000 AND not inside a loop
• The table is clearly a small config/lookup table
• Query total cost is already very low (< 100)

For YugabyteDB: Lower the threshold to 1,000 rows for 🔴 Critical. LSM-tree storage makes sequential scans significantly more expensive than in standard Postgres due to SST file structure. Mention this explicitly.

What to Report for Each Significant Seq Scan

1. Table name and estimated / actual row count
2. Filter condition — this is your index candidate
3. Rows removed by filter if present — high discard = strong index opportunity
4. Loop count if inside a Nested Loop
5. Specific index recommendation based on the filter

Index recommendation format:

-- If filter is: (status = 'pending')
CREATE INDEX ON orders (status);

-- If filter is: (user_id = $1 AND created_at > $2)
CREATE INDEX ON orders (user_id, created_at);

-- If most rows have status='completed' and you only query 'pending':
CREATE INDEX ON orders (status) WHERE status != 'completed';

Stage 4: Secondary Issue Checks

After seq scans, check for these in order of typical impact:

4a. Row Estimate Mismatch (ANALYZE only)

• If estimated rows vs actual rows differ by > 10×, flag stale or bad statistics
• Common cause: recent bulk loads, heavy deletes, or correlated columns
• Recommendation: ANALYZE table_name; or CREATE STATISTICS for correlated columns
• Very large mismatches (> 100×) can cause the planner to choose completely wrong join strategies

4b. Nested Loop with Large Row Counts

• A Nested Loop is efficient for small outer sets (< ~1,000 rows)
• If outer side has many rows AND inner side does a Seq Scan or expensive lookup: flag it
• The actual cost is: outer_rows × inner_cost_per_loop
• Recommendation: consider forcing a Hash Join via SET enable_nestloop = off temporarily to test, or add an index on the join column

4c. Hash Join Spilling to Disk

• Visible as: Batches: N where N > 1 in a Hash node
• Means the hash table didn't fit in work_mem
• Recommendation: SET work_mem = '64MB'; (or higher, test in session first)

4d. Sort Spilling to Disk

• Visible as: Sort Method: external merge or external sort
• Recommendation: increase work_mem

4e. High "Rows Removed by Filter" on an Index Scan

• Means the index is finding rows but many are discarded by a post-scan filter
• The index may not be selective enough, or a composite index with the filter column would help

4f. Bitmap Heap Scan with High Recheck

• Rows Removed by Index Recheck: N being high means lossy bitmap pages
• Recommendation: increase work_mem so the bitmap stays exact

4g. Very High Total Cost / Slow Execution

• If EXPLAIN ANALYZE shows total actual time > 1,000ms for what should be a simple query, flag it even if no single obvious issue dominates

Stage 5: Output Format

Structure your response exactly as follows:

Plan Type

State whether this is EXPLAIN, EXPLAIN ANALYZE, or EXPLAIN (ANALYZE, BUFFERS), and note any limitations this places on the analysis.

Plan Overview

1–2 sentences: what the query does, overall shape (joins, aggregations, etc.), and total cost or execution time if available.

🔴 Critical Issues

One section per issue. For each:

• Issue title (e.g., "Seq Scan on orders — 450,000 rows")
• What it is and why it matters
• Specific fix with SQL if applicable

If no critical issues: say "No critical issues found."

🟡 Secondary Issues

Brief bullets for lower-priority findings (row estimate mismatches, memory spills, etc.)

If none: omit this section.

✅ Recommended Actions

Numbered list, ordered by expected impact:

1. Most impactful fix (usually the biggest seq scan index)
2. Second fix
3. ...

Keep this list actionable — real SQL commands where possible.

Guidelines

• Always explain why something is a problem, not just that it is
• Be specific: name the table, show the filter, write the actual CREATE INDEX statement
• If the plan is truncated, say so and analyze what's available
• Do not recommend indexes that are already being used (already showing Index Scan in the plan)
• If the plan looks healthy, say so clearly — a good plan is worth confirming
• For EXPLAIN-only (no ANALYZE): caveat your findings with "based on planner estimates"
• If the user mentions this is YugabyteDB, apply the lower thresholds and add YB-specific context