agentleFS
Sign inStart free
jdanigojdanigo/hydraia/skills/db-optimization/SKILL.md

db-optimization

Skill8 starsChanged 2 months agoMITView on GitHub

Use when the bottleneck is database-shaped — slow queries, lock timeouts, connection exhaustion, N+1 patterns, index questions, schema migrations on hot tables. Evidence-first DB playbook for PostgreSQL, MySQL/MariaDB, MongoDB, and Redis backing the db-performance-tuner agent.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 4 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read-first: no change is proposed without its EXPLAIN plan, stat, or code evidence.

    line 11
  2. Distinctive:

    All write-shaped operations (create index, schema change, config change) are PLAN TASKS — reviewed, gated, never executed by a diagnosis …

    line 12
  3. Very specific:

    Expand: add the new column/table/index alongside the old — additive, backward-compatible, online.

    line 20
  4. Distinctive:

    Contract: only after a full bake period, remove the old shape in a later release.

    line 23
  5. Distinctive:

    Always set: acquisition timeout, idle timeout, max lifetime.

    line 41

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

DB Optimization — EXPLAIN before you touch anything

Playbook for the DB half of the Hydraia perf route. The db-performance-tuner agent does the diagnosis; this skill sets the rules the route runs under.

<HARD-RULES> - Read-first: no change is proposed without its EXPLAIN plan, stat, or code evidence. Live DB unavailable → code/schema inference only, tagged UNVERIFIED, with the confirming commands listed for the human. - All write-shaped operations (create index, schema change, config change) are PLAN TASKS — reviewed, gated, never executed by a diagnosis agent. - Destructive or blocking operations (table rewrites, long locks) are flagged for explicit human approval in the plan; prefer and name the online variant per engine: `CREATE INDEX CONCURRENTLY` (Postgres), `ALGORITHM=INPLACE, LOCK=NONE` (MySQL), rolling index builds (Mongo). - Schema changes on live tables follow EXPAND-CONTRACT, no exceptions. - Redact literal data values from any plan/log excerpt that lands in a committed artifact. </HARD-RULES>

Expand-contract migration protocol

  1. Expand: add the new column/table/index alongside the old — additive, backward-compatible, online.
  2. Dual-write + backfill: application writes both shapes; backfill history in batches (bounded, throttled, resumable).
  3. Switch reads to the new shape behind a verified flag; watch metrics.
  4. Contract: only after a full bake period, remove the old shape in a later release. Contract is its own plan task, never bundled with expand.

Index design guidance

  • Composite order: equality columns first, then range, then sort — match the query, not intuition.
  • Covering indexes to eliminate lookups for hot read paths; weigh write amplification (every index taxes every write).
  • Partial/filtered indexes for skewed predicates (WHERE deleted_at IS NULL).
  • Hunt UNUSED indexes too (pg_stat_user_indexes, sys.schema_unused_indexes) — dropping one is also an optimization.
  • The index is not always the answer: sometimes it is the query shape, the schema, or the caller (N+1, missing cache).

N+1 patterns per ORM family

  • Detection: a loop in code + per-iteration query in the log/profile. The agent cites the code location.
  • Fix families: eager loading (include/select_related/prefetch_related/preload), query batching (IN lists), dataloader pattern (GraphQL), denormalized read models for hot aggregates.

Pooling guidance

  • Pool size starts near connections = cores * 2 + effective_spindles (HikariCP heuristic) — not hundreds; queueing beats connection thrash.
  • Always set: acquisition timeout, idle timeout, max lifetime. Connection churn (open/close per request) is itself a finding.

Per-engine evidence commands

See the db-performance-tuner agent definition for the full read-only evidence list per engine (EXPLAIN variants, stat views, lock views, SLOWLOG). This skill and that agent must stay in sync — update both or neither.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free