Database Code One Task Group

This documentation was generated with the assistance of AI. Please report any inaccuracies.

Implement every task in one database-tagged plan-group bucket by calling iru-database-code-one-task once per task, one at a time, then bring implementation_plan.md up to date and notify the user of the bucket’s outcome.

Purpose

iru-database-code-one-task-group is the database counterpart to iru-java-code-one-task-group/ iru-dotnet-code-one-task-group, but deliberately simpler. It never dispatches tasks in parallel, even when the bucket is marked Parallelizable: yes — database changes (schema migrations, seed data, index changes, and the queries a task calls for) commonly carry a real ordering dependency even when a plan doesn’t flag one explicitly. It also captures no pre-change quality baseline and runs no bucket-wide test/coverage/quality validation pass of its own, since there is no single standardized toolchain across database platforms the way Maven/dotnet provide for Java/.NET — it relies on iru-database-code-one-task to generate (and, where safe, execute) each task’s queries and record them in database-report.md.

Purpose group

Database Development Tooling

Invocation

/iru-database-code-one-task-group <bucket text>

Inputs

Argument Required Description Default

bucket text

Yes

Every task/sub-task in the bucket, each with its own text, plus any relevant "Current code state" context.

None — this skill has nothing to do without a bucket to implement.

Outputs

  • Each task in the bucket implemented via iru-database-code-one-task, one at a time, in the plan’s order — queries generated and, where a matching database MCP is connected, the read-only ones executed.

  • database-report.md at the repository root, appended to once per task, with a standing reminder to verify its contents and that whether to commit it is a decision for a later step.

  • implementation_plan.md updated once every task has been attempted: each completed task’s (and its sub-tasks') checkbox checked, with a short note naming the queries generated/executed; any blocked task left unchecked with its blocker noted instead.

  • A user-facing notification summarizing the bucket’s outcome — which tasks completed, and which (if any) are blocked and why — repeating the database-report.md verify/commit reminder if any task touched it.

  • A summary handed back to the caller (iru-code-one-task-group): per task, the queries generated/executed and whether it finished or blocked.

  • No pre-change quality baseline, and no bucket-wide test/coverage/quality validation — unlike iru-java-code-one-task-group/ iru-dotnet-code-one-task-group.

Execution flow

flowchart TD A["Start /database-code-one-task-group\n(bucket text)"] --> B["For each task, in plan order:\ninvoke database-code-one-task"] B --> C{Task reported blocked?} C -- yes --> D["Record blocker;\ncontinue to next task"] C -- no --> E["Record queries generated/executed\n+ database-report.md updated;\ncontinue to next task"] D --> F{More tasks in bucket?} E --> F F -- yes --> B F -- no --> G["Check off completed tasks'\ncheckboxes in implementation_plan.md;\nnote blockers for the rest"] G --> H["Notify user of bucket outcome"] H --> I["Report bucket outcome to caller\n(code-one-task-group)"]

Dependencies

Invokes

  • Database Code One Task — once per task in the bucket, one at a time, in the plan’s order — generates (and, where safe, executes) that task’s queries and records them in database-report.md.

Invoked by

  • Code One Task Group — dispatches every task-group bucket tagged (or defaulted to) database to this skill, once per bucket.

Related agents

None — iru-database-code-one-task-group calls iru-database-code-one-task directly via the Skill tool rather than through a sub-agent.

Source

SKILL.md on GitHub — the file this page was generated from.