Skip to main content
Glama

MySQL AIops

Audit availability: Real writes require a committed pending audit row and stop before contacting the platform if storage is unavailable. Reads and previews may continue with a warning; new audit rows store outcome summaries rather than full responses.

Disclaimer: Community-maintained open-source project. Not affiliated with, endorsed by, or sponsored by Oracle Corporation or the MariaDB Foundation. "MySQL" is a trademark of Oracle Corporation; "MariaDB" is a trademark of MariaDB plc; all product/trademark names belong to their respective owners. MIT licensed.

Governed AI-ops for MySQL 8.x and MariaDB 10.6+ DBA operations — connecting to a server with PyMySQL and reading information_schema / performance_schema — with a built-in governance harness: unified audit log, token/runaway budget guard, undo-token recording, and descriptive risk-tier labels. The server flavor (mysql vs mariadb) is detected from version() and flavor-dependent statements branch automatically (SHOW REPLICA STATUS vs SHOW SLAVE STATUS).

What it does

Four flagship signature analyses, plus the guarded reads and writes around them:

  • Slow-query RCA — take the worst events_statements_summary_by_digest entry (plus an optional EXPLAIN FORMAT=JSON plan) and map its numbers — no-index share (SUM_NO_INDEX_USED), lock-time share, rows-examined/sent ratio, tmp-disk spill, call count, plan access types — to a cited cause and a concrete action. Every finding carries its measured number, not a black-box verdict.

  • InnoDB lock-wait & deadlock chain RCA — build the wait-for tree from performance_schema.data_lock_waits (MariaDB: information_schema.innodb_lock_waits), name the root blocker (blocks others, waits on none), and parse the last deadlock out of SHOW ENGINE INNODB STATUS.

  • Replication lag RCA — map the replica's IO/SQL thread state, Seconds_Behind_Source and error fields to a cited cause + action (stopped IO thread, failed applier statement, lagging applier, intentional SQL_Delay).

  • Table fragmentation analysis — rank tables by reclaimable data_free from information_schema.tables into cited OPTIMIZE TABLE candidates.

Related MCP server: postgres-aiops

What works

  • CLI (mysql-aiops ...): init, overview, server, activity, query, index, table, repl, analyze, remediate, secret, doctor, mcp.

  • MCP server (mysql-aiops mcp or mysql-aiops-mcp): 35 tools (26 read, 9 write), every one wrapped with the bundled @governed_tool harness.

  • Encrypted credentials: the account password lives in an encrypted store ~/.mysql-aiops/secrets.enc (Fernet + scrypt) — never plaintext on disk. Unlock with a master password from MYSQL_AIOPS_MASTER_PASSWORD (MCP/CI) or an interactive prompt (CLI).

  • Reversibility: mutating writes fetch the real before-state first and record a faithful inverse — create_index↔drop_index; drop_index captures the index definition out of SHOW CREATE TABLE so undo recreates it exactly; set_global_variable captures the prior value from SHOW GLOBAL VARIABLES so undo sets it back. Irreversible ops (kill_session, kill_query, optimize_table, analyze_table, reset_query_stats) record prior state for audit but declare no undo.

  • Safety: every state-changing CLI op supports --dry-run and requires double confirmation; every write MCP tool takes a dry_run preview. All identifiers that cannot be parameterised (schema/table/index/column/variable names) are validated against a strict charset and backtick-quoted; all values are bound query parameters.

What this tool does, and does not, decide

It delivers MySQL / MariaDB DBA operations — reads and writes — accurately and efficiently, and records every one of them. It does not decide whether a write is allowed to happen. That is the agent's judgement, or the permission of the account you connect it with: point it at a MySQL/MariaDB account granted only SELECT / PROCESS / REPLICATION CLIENT and no write privileges (no INSERT/UPDATE/DELETE/DDL), and the writes fail at the server — the place that actually owns the permission.

So there is no read-only switch, no policy file, no approval gate to configure. The one thing the tool guarantees is that nothing is silent: every call, over MCP and over the CLI alike, lands an audit row in ~/.mysql-aiops/audit.db, and destructive writes still capture their before-state and record an inverse where one exists.

Each tool declares a risk_level, carried into the audit row as a descriptive tier (none/confirm/review) — so a reviewer can see at a glance that a row was a high-risk delete. It is a label, not a gate.

Running a smaller / local model? See agent-guardrails.md — it lists the guardrails this tool now enforces for you (so you don't spend prompt budget restating them) and gives a ready-made system prompt for what's left.

Capability matrix (35 MCP tools)

Domain

Tools

Count

R/W

Overview

overview

1

read

Server

server_version, show_variables, show_status, list_databases, list_engines, connection_stats

6

read

Activity

list_sessions, long_running_queries, list_transactions, lock_waits

4

read

Queries

top_queries, explain_query

2

read

Indexes

unused_indexes, redundant_indexes, index_stats

3

read

Tables

table_sizes, table_fragmentation, table_status

3

read

Replication

replica_status, binlog_status

2

read

Analysis (flagship)

slow_query_rca, lock_wait_rca, replication_lag_rca, fragmentation_analysis

4

read

Writes

kill_session, kill_query, drop_index

3

write (high)

optimize_table, analyze_table, create_index, set_global_variable, reset_query_stats

5

write (medium)

Undo

undo_list, undo_apply

2

read / write

The flagship analyses accept injected records for pure/offline analysis, or pull live from a configured target. top_queries/slow_query_rca require performance_schema=ON; the read account should have PROCESS, REPLICATION CLIENT and SELECT on performance_schema.

Support scope

Platform

Status

MySQL 8.0 / 8.4

targeted (SHOW REPLICA STATUS, performance_schema.data_lock_waits)

MariaDB 10.6+ / 11.x

targeted (SHOW SLAVE STATUS, information_schema.innodb_lock_waits)

MySQL 5.7 and older

not targeted (EOL; pre-8.0 digest/lock views untested)

Cloud-managed MySQL (RDS/Aurora/Cloud SQL flavors)

wire-compatible reads should work; managed restrictions (KILL, SET GLOBAL) apply — untested

Quick start

As a Claude Code plugin

One install gives an agent both the skill and the MCP server:

/plugin marketplace add AIops-tools/marketplace
/plugin install mysql-aiops@aiops-tools

The MCP server is fetched with uv and pinned to the package version this plugin declares, so an audit row can be traced back to the code that wrote it. Credentials are still configured with mysql-aiops init — see below.

As an OpenClaw plugin

The same bundle is published on ClawHub, where one install delivers the skill and its MCP server together:

openclaw plugins install clawhub:@zw008/mysql-aiops
openclaw skills info mysql-aiops          # expect: Visible to model: yes

Restart the OpenClaw gateway afterwards so it loads the plugin. The MCP server is fetched with uv, pinned to this exact release, so uvx has to be on PATH — without it the skill still installs but reports Visible to model: no. Credentials are configured exactly as below.

As a CLI or standalone MCP server

uv tool install mysql-aiops             # or: pipx install mysql-aiops
mysql-aiops init                        # wizard: add a target + store the password (encrypted)
mysql-aiops doctor                      # verify config, secrets, connectivity, flavor, perf-schema
mysql-aiops overview                    # one-shot server health snapshot
mysql-aiops analyze slow-query          # RCA the worst statement digest
mysql-aiops analyze fragmentation       # OPTIMIZE TABLE candidates

Run as an MCP server (stdio):

export MYSQL_AIOPS_MASTER_PASSWORD=...  # unlock secrets non-interactively
mysql-aiops-mcp

Where that password then lives: an exported variable is readable by every process this shell starts and is recorded by shell history. On a shared or long-lived host, prefer the interactive prompt, or inject it from a secret manager for the life of the one command that needs it.

Claude Desktop / MCP client config:

{
  "mcpServers": {
    "mysql-aiops": {
      "command": "uvx",
      "args": ["--from", "mysql-aiops", "mysql-aiops-mcp"],
      "env": { "MYSQL_AIOPS_MASTER_PASSWORD": "your-master-password" }
    }
  }
}

Env-block caveat: the env block above is the only environment the MCP server sees — GUI-launched clients do not inherit your shell profile. Put MYSQL_AIOPS_MASTER_PASSWORD (and MYSQL_AIOPS_HOME / MYSQL_AUDIT_APPROVED_BY if you use them) there, or the server cannot unlock the secret store.

Governance

Every MCP tool passes through the bundled @governed_tool harness:

  • Audit — every call (params, outcome summary, status, duration, risk tier, and any operator-supplied approver/rationale) is logged to ~/.mysql-aiops/audit.db (relocatable via MYSQL_AIOPS_HOME). The CLI writes the same row the MCP path does — there is no unaudited entry point.

  • Runaway guard — a safety backstop, not an authorization gate: the same call hammered in a tight loop trips a circuit breaker. Disable with MYSQL_RUNAWAY_MAX=0; optional hard ceilings via MYSQL_MAX_TOOL_CALLS / MYSQL_MAX_TOOL_SECONDS.

  • Undo recording — reversible writes record an inverse descriptor built from the fetched before-state.

  • Risk tier — a descriptive label on the audit row derived from risk_level; it gates nothing.

Scope

This is the MySQL / MariaDB DBA-ops member of the AIops-tools family (governed AI-ops with audit + budget + undo + risk tiers). Do NOT use it for PostgreSQL — use postgres-aiops. Do NOT use it for OT / industrial edge — see the separate industrial-aiops line — nor for application-schema migrations or ORM management.

Missing a capability?

Coverage is intentionally a curated subset of MySQL's catalogs and maintenance surface. Missing a view, a metric, or a maintenance command? Open an issue or PR — contributions welcome. 缺功能提 issue/PR 欢迎留言。

Verification status

Live-verified against MySQL 8.4.10 and MariaDB 11.8.8 (2026-07-19/20). Connectivity, the reads, analyze slow-query on genuine full scans, and the full governance loop (real create_index → audit row → undo actually dropping it) were exercised against a real server. That run found and fixed a real bug the mock suite could not see: SUM() aggregates come back as Decimal, which is not JSON serializable.

The MariaDB branch is now verified too, including analyze lock-waits against real row contention (it correctly identified the root blocker and the measured wait). Replication against a real replica remains unverified.

docs/VERIFICATION.md records exactly what was checked and what is still open. mysql-aiops doctor is the fastest live check (connectivity, flavor, performance_schema, replica role).

Available Tools

35 tools
analyze_tableA
Destructive

[WRITE][risk=medium] ANALYZE TABLE to refresh index statistics.

No undo; captures prior stats for audit. Takes a brief read lock while sampling. Pass dry_run=True to preview.

Args: table: Table name (optionally schema-qualified). dry_run: If True, preview without running. target: Target name from config; omit for the default.

Inputs: table (required), target, dry_run.

binlog_statusB
Read-only

[READ] Binary-log configuration, GTID mode, and connected downstream replicas.

Args: target: Target name from config; omit for the default.

Inputs: target.

connection_statsA
Read-only

[READ] Connection counters vs max_connections (exhaustion early warning).

Args: target: Target name from config; omit for the default.

Inputs: target.

create_indexA
Destructive

[WRITE][risk=medium] Create an index. Reversible: undo drops the created index.

The created (table, name) is returned so the harness records an undo that drops exactly this index. Pass dry_run=True to preview. Alternatively pass definition (a captured CREATE INDEX statement — this is how drop_index's undo descriptor replays) INSTEAD of table/columns.

Args: table: Table to index (optionally schema-qualified). Required unless definition is given. columns: Column names to index. Required unless definition is given. name: Index name (auto-generated from table+columns when omitted). unique: Create a UNIQUE index. definition: A full CREATE [UNIQUE] INDEX statement to execute verbatim (shape-validated). Mutually exclusive with table/columns. dry_run: If True, preview without creating. target: Target name from config; omit for the default.

Inputs: name, table, target, unique, +3 more.

drop_indexA
Destructive

[WRITE][risk=high] Drop an index. Reversible: captures its definition first.

Before dropping, the exact index definition is rebuilt from SHOW CREATE TABLE so the harness records an undo that recreates it. Pass dry_run=True to preview.

Args: table: Table the index belongs to (optionally schema-qualified). name: Index name to drop. dry_run: If True, preview without dropping. target: Target name from config; omit for the default.

Inputs: name (required), table (required), target, dry_run.

explain_queryA
Read-only

[READ] Return the JSON execution plan for sql (EXPLAIN FORMAT=JSON).

The statement is planned, not executed.

Args: sql: A single SQL statement to EXPLAIN. target: Target name from config; omit for the default.

Inputs: sql (required), target.

fragmentation_analysisA
Read-only

[READ] Rank tables by reclaimable data_free into OPTIMIZE TABLE candidates.

Pass 'tables' (as from table_fragmentation) for pure/offline analysis, or omit to pull the worst-fragmented tables live. Each recommendation cites its numbers.

Args: tables: Injected fragmentation rows; if omitted, pulled live. limit: How many tables to pull when not injected (default 50). target: Target name from config; omit for the default.

Inputs: limit, tables, target.

index_statsA
Read-only

[READ] Per-index column lists and cardinality (selectivity screening).

Args: target: Target name from config; omit for the default.

Inputs: target.

kill_queryA
Destructive

[WRITE][risk=high] Cancel a session's running statement (KILL QUERY). No inverse.

The session stays connected; only its current statement is aborted. Captures the session's user/host/query for audit. Pass dry_run=True to preview.

Refuses this tool's own session id — the statement it would cancel is this very call. Enforced under dry_run too.

Args: session_id: Session/connection id (from list_sessions). dry_run: If True, preview without cancelling. target: Target name from config; omit for the default.

Inputs: session_id (required), target, dry_run.

kill_sessionA
Destructive

[WRITE][risk=high] Terminate a session (KILL CONNECTION). No safe inverse.

Captures the session's user/host/query for the audit trail; a kill cannot be undone, so no undo is offered. Pass dry_run=True to preview.

Refuses this tool's own session id — including under dry_run, which must report a refusal rather than preview a call that will be refused.

Args: session_id: Session/connection id (from list_sessions). dry_run: If True, preview without killing. target: Target name from config; omit for the default.

Inputs: session_id (required), target, dry_run.

list_databasesA
Read-only

[READ] User schemas with table count and data/index size (largest first).

Args: target: Target name from config; omit for the default.

Inputs: target.

list_enginesA
Read-only

[READ] Storage engines and which is the default (SHOW ENGINES).

Args: target: Target name from config; omit for the default.

Inputs: target.

list_sessionsA
Read-only

[READ] Current sessions (processlist) with per-command counts.

Flags sleeping sessions (connection-pool leak suspects).

Args: include_sleeping: Include sessions in command=Sleep (default True). target: Target name from config; omit for the default.

Inputs: target, include_sleeping.

list_transactionsA
Read-only

[READ] Open InnoDB transactions, oldest first (stuck-transaction hunting).

Args: target: Target name from config; omit for the default.

Inputs: target.

lock_wait_rcaA
Read-only

[READ] Build the InnoDB wait-for tree, name the root blocker, and parse the last deadlock out of SHOW ENGINE INNODB STATUS.

Pass 'pairs' (as from lock_waits) for pure/offline analysis, or omit to pull the current lock-wait graph live.

Args: pairs: Injected lock-wait pairs {blockedId, blockingId, ...}; if omitted, pulled live (flavor-branched). include_deadlock: Also parse the LATEST DETECTED DEADLOCK section from SHOW ENGINE INNODB STATUS (default True). target: Target name from config; omit for the default.

Inputs: pairs, target, include_deadlock.

lock_waitsA
Read-only

[READ] InnoDB wait-for edges (blocked session -> blocking session).

Reads performance_schema.data_lock_waits on MySQL 8.x, or information_schema.innodb_lock_waits on MariaDB (flavor-branched).

Args: target: Target name from config; omit for the default.

Inputs: target.

long_running_queriesA
Read-only

[READ] Active statements running at least min_seconds, oldest first.

Args: min_seconds: Minimum age in seconds (default 60). target: Target name from config; omit for the default.

Inputs: target, min_seconds.

optimize_tableA
Destructive

[WRITE][risk=medium] OPTIMIZE TABLE (rebuild, reclaim data_free). Records prior stats.

No undo (a rebuild has no inverse); the prior size/fragmentation stats are captured for the audit trail. InnoDB maps this to ALTER TABLE ... FORCE (online DDL, brief locks) — schedule off-peak for hot tables. Pass dry_run=True to preview.

Args: table: Table name (optionally schema-qualified, e.g. shop.orders). dry_run: If True, preview without running. target: Target name from config; omit for the default.

Inputs: table (required), target, dry_run.

overviewA
Read-only

[READ] One-shot server health snapshot.

Version + flavor (mysql/mariadb) + uptime, connection headroom vs max_connections, sessions by command, the longest-running query, the most fragmented table, and the replica role — each section captured defensively so one failing probe does not sink the rest.

Args: target: Target name from config; omit for the default.

Inputs: target.

redundant_indexesB
Read-only

[READ] Indexes whose columns are a leading prefix of another index (dupes).

Args: target: Target name from config; omit for the default.

Inputs: target.

replica_statusA
Read-only

[READ] Replica thread state and lag (SHOW REPLICA/SLAVE STATUS, flavor-branched).

Empty on a primary/standalone server. MySQL 8.x uses SHOW REPLICA STATUS; MariaDB uses SHOW SLAVE STATUS — the result is normalised either way.

Args: target: Target name from config; omit for the default.

Inputs: target.

replication_lag_rcaA
Read-only

[READ] Map replica thread state / lag / error fields to cause + action.

Pass 'status' (as from replica_status) for pure/offline analysis, or omit to pull the live replica status (flavor-branched).

Args: status: Injected replica status record; if omitted, pulled live. target: Target name from config; omit for the default.

Inputs: status, target.

reset_query_statsA
Destructive

[WRITE][risk=medium] Reset statement-digest accumulators (irreversible).

Truncates performance_schema.events_statements_summary_by_digest; the counters cannot be restored, so no undo is recorded. Pass dry_run=True to preview.

Args: dry_run: If True, preview without resetting. target: Target name from config; omit for the default.

Inputs: target, dry_run.

server_versionB
Read-only

[READ] Server version, flavor (mysql/mariadb), uptime and read-only state.

Args: target: Target name from config; omit for the default.

Inputs: target.

set_global_variableA
Destructive

[WRITE][risk=medium] SET GLOBAL a server variable. Reversible: captures prior value.

Runtime-only: the change does not survive a restart (persist it in my.cnf — or SET PERSIST on MySQL 8 — yourself); reported but NOT performed automatically. The prior value (from SHOW GLOBAL VARIABLES) is captured so the harness records an undo that sets it back. Pass dry_run=True to preview.

Refuses the globals that would lock this tool out of the server (init_connect, max_connections, max_user_connections, read_only, super_read_only, skip_networking, require_secure_transport, and wait_timeout/interactive_timeout below 30s). SET GLOBAL is immediate, so those strand the undo. Enforced under dry_run too.

Args: name: The global variable name (e.g. max_connections). value: The new value (as a string). dry_run: If True, preview without changing. target: Target name from config; omit for the default.

Inputs: name (required), value (required), target, dry_run.

show_statusB
Read-only

[READ] Global status counters (SHOW GLOBAL STATUS).

Args: pattern: Optional substring to filter counter names (e.g. 'Threads', 'Innodb_buffer_pool'). target: Target name from config; omit for the default.

Inputs: target, pattern.

show_variablesA
Read-only

[READ] Global variables (SHOW GLOBAL VARIABLES).

Args: pattern: Optional substring to filter variable names (e.g. 'innodb_buffer', 'max_connections'). target: Target name from config; omit for the default.

Inputs: target, pattern.

slow_query_rcaA
Read-only

[READ] RCA for the worst statement digest, with cause + action.

Picks the digest with the greatest total time and maps its numbers (no-index share, lock-time share, examined/sent ratio, tmp-disk spill, calls) — plus an optional EXPLAIN plan — to cited causes and concrete actions. Pass 'statements' for pure/offline analysis, or omit to pull the top digests live from performance_schema.

Args: statements: Injected digest rows (as from top_queries); if omitted, the worst statements are pulled live. explain_sql: Optional SQL to EXPLAIN so plan access types feed the RCA. limit: How many statements to pull when not injected (default 20). target: Target name from config; omit for the default.

Inputs: limit, target, statements, explain_sql.

table_fragmentationA
Read-only

[READ] data_free per table (space OPTIMIZE TABLE could reclaim), worst first.

Args: limit: Number of tables to inspect (default 50). target: Target name from config; omit for the default.

Returns an envelope: {"tables": [...], "returned": N, "limit": L, "truncated": bool}. When "truncated" is true there is more beyond what was returned — re-run with a higher limit rather than treating this as the complete picture.

Inputs: limit, target.

table_sizesA
Read-only

[READ] Largest tables by data + index size.

Args: limit: Number of tables to return, largest first (default 20). target: Target name from config; omit for the default.

Returns an envelope: {"tables": [...], "returned": N, "limit": L, "truncated": bool}. When "truncated" is true there is more beyond what was returned — re-run with a higher limit rather than treating this as the complete picture.

Inputs: limit, target.

table_statusA
Read-only

[READ] Per-table engine, row format, row estimate and last update time.

Flags non-InnoDB tables (no row-level locking / crash recovery).

Args: limit: Number of tables to inspect (default 50). target: Target name from config; omit for the default.

Returns an envelope: {"tables": [...], "returned": N, "limit": L, "truncated": bool}. When "truncated" is true there is more beyond what was returned — re-run with a higher limit rather than treating this as the complete picture.

Inputs: limit, target.

top_queriesA
Read-only

[READ] Top statement digests from performance_schema by a whitelisted metric.

Args: order_by: One of total_time, mean_time, calls, rows_examined, lock_time, no_index. limit: Number of statements to return (1..200, default 20). target: Target name from config; omit for the default.

Returns an envelope: {"statements": [...], "returned": N, "limit": L, "truncated": bool}. When "truncated" is true there is more beyond what was returned — re-run with a higher limit rather than treating this as the complete picture.

Inputs: limit, target, order_by.

undo_applyA
Destructive

[WRITE][risk=medium] Apply a recorded undo by dispatching its inverse tool.

The inverse runs through its own governed tool, so its own risk tier and audit row are recorded there. Pass dry_run=True to preview the inverse call without executing it. A token can only be applied once.

Args: undo_id: The undoId from undo_list (or an _undo_id in a write result). dry_run: If True, preview the inverse tool + params without running it. target: Passed through to the inverse tool when it accepts a target.

Inputs: undo_id (required), target, dry_run.

undo_listA
Read-only

[READ] List recorded, not-yet-applied undo tokens (most recent first).

Each entry names the original tool, the inverse tool that undo_apply would run, and a human note. Use the undoId with undo_apply.

Returns {"undos": [...], "returned": N, "limit": L, "truncated": bool}. truncated is measured (one extra row is fetched), not guessed from a length coincidence: when it is true there are MORE tokens than shown, so re-run with a higher limit rather than reporting the list as complete.

Each entry carries effectVerified. False means the original write lost its response, so the change it reverses is PROBABLE, not confirmed — check the live state before applying, and do not report the result as a restore of a state that may never have been reached.

Args: limit: Max rows to return (default 50, capped at 500). target: Unused (undo state is host-local); accepted for CLI uniformity.

Inputs: limit, target.

unused_indexesA
Read-only

[READ] Secondary indexes with zero I/O events since restart (drop candidates).

From performance_schema.table_io_waits_summary_by_index_usage — counters reset on server restart, so confirm over a full business cycle.

Args: target: Target name from config; omit for the default.

Inputs: target.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A3.7/5.0

    Scored across 35 tools

    Disambiguation4/5

    Most tools have clearly distinct purposes: kill_session vs kill_query, table_sizes vs table_fragmentation vs table_status, and the read/analysis pairs (lock_waits vs lock_wait_rca, top_queries vs slow_query_rca) are well separated, with descriptions explicitly cross-referencing each other. The main overlap is the one-shot 'overview' tool, which duplicates subsets of server_version, connection_stats, long_running_queries and table_fragmentation, creating mild redundancy an agent could stumble over.

    Naming Consistency4/5

    Nearly all names are snake_case and readable, with a sensible convention of verb_noun for actions (kill_session, create_index, set_global_variable) and noun-phrase for reads (table_sizes, index_stats). Minor deviations exist (undo_list vs undo_apply, the single-word 'overview', and the 'rca' suffix group), but they are consistent within their families.

    Tool Count3/5

    35 tools is on the heavy side for a single server and sits above the typical 3-15 range; several read probes (server_version, connection_stats, show_status) are partially subsumed by 'overview'. However, the breadth of the MySQL AIOps domain (health, sessions, locks, replication, indexes, fragmentation, queries, plus governed writes) means most tools target a genuinely distinct signal rather than padding.

    Completeness4/5

    Coverage is strong and coherent: a full read surface (status, variables, sessions, transactions, lock-wait/slow-query/replication RCA, index and fragmentation analysis, replication/binlog state) plus reversible writes (kill, optimize/analyze, create/drop index, set global variable, reset stats) backed by an undo framework. Gaps are minor for the AIOps scope: no replication-control actions (start/stop/promote replica) and no schema/table DDL beyond index operations.

    Maintenance

    ActivityMaintained
    ResponsivenessOutdated data

    Related MCP Connectors

    Related MCP Servers

    • F
      license
      Not graded
      quality
      C
      maintenance
      Universal database copilot for diagnostics, operations, and performance analysis via MCP and CLI.
      -
    • A
      license
      A
      quality
      B
      maintenance
      Governed PostgreSQL DBA operations — slow-query, bloat, and blocking-lock RCA, index management, vacuum/analyze, and replication lag, with unbypassable audit logging (MCP + CLI), budget/runaway guards, dry-run, and undo/rollback.
      35
      MIT
    • A
      license
      A
      quality
      B
      maintenance
      Governed Prometheus + Grafana operations — firing-alert and scrape-target RCA, alert noise/flapping analysis, silences, and dashboards, with unbypassable audit logging (MCP + CLI), budget/runaway guards, dry-run, and undo/rollback.
      39
      MIT
    • A
      license
      A
      quality
      B
      maintenance
      Governed OPNsense + pfSense firewall operations — gateway-health, rule-shadow, and blocked-traffic RCA, with guarded rule/alias writes, unbypassable audit logging (MCP + CLI), budget/runaway guards, dry-run, and undo/rollback.
      35
      2
      MIT