How to implement Db2 migration support
The source reports Db2 support completed in less than two days with twenty-seven merged PRs and test coverage equivalent to PostgreSQL; two escalations required a human.
cockroachlabs.comSuggested prompt
Suggested prompt written by Agents In Use, not the source author. Replace bracketed inputs and review tool permissions first.
Implement [source database] support in [migration tools] from [GitHub issue]. Decompose the work into dependent issues for types, row iteration, Fetch, Verify, Convert, CI and fixtures. Review plans before coding, split oversized issues, file discovered gaps, require tests and reviews before merging, and escalate [defined blockers] to a human.
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- rafiss
- Original account
- Read the source ↗
- Published
- Source link checked
- Reported result
- The source reports Db2 support completed in less than two days with twenty-seven merged PRs and test coverage equivalent to PostgreSQL; two escalations required a human.
- Human involvement
- The pipeline required planning and review handoffs; two of nine escalated issues reached humans, and the source says no human wrote the code.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Open a GitHub issue describing Db2 migration support.
- Decompose the request into dependency-ordered sub-issues.
- Split oversized workups and file issues for discovered defects.
Cockroach Labs describes using its MOLT Sinai coding pipeline to add IBM Db2 source support to its database migration tools. The work included a schema converter, Fetch and Verify paths, a vendored ANTLR grammar, a CI Docker image, and more than ten thousand lines of test fixtures. The supplied excerpt identifies the pipeline but does not name its underlying agent model.
The process began with a GitHub issue describing the request. A planning agent split it into fifteen dependent sub-issues, ordered from foundation and types through row iteration, Fetch, Verify, Convert, CI, and test data. Two issues were divided further after their own workups found them too large. Agents also filed another dozen issues about their work, including fixture gaps, a type-mapping bug, and an isolation-level correction.
The source reports thirty-two sub-issues by closure, twenty-seven merged pull requests, fifty-five reviewer send-backs, and nine escalations, two to a human. It says coverage matched PostgreSQL, its best-tested dialect, and that no human wrote the code. The reported Db2 effort ran from a Wednesday evening to Friday afternoon. Who it's for: migration-tool teams that can define dependency plans, require reviews and tests, and handle explicit escalations before merging changes.
Prompt
Suggested prompt (written by Agents In Use, not the author's)
Source: Blog · rafiss
Frequently asked questions
How did the agents divide the Db2 request?
A planning agent created fifteen sub-issues with an explicit dependency graph. Two were decomposed again, and agents filed additional issues about gaps and defects in their own work.
What review and escalation results were reported?
The parent effort ended with twenty-seven merged PRs, fifty-five reviewer send-backs and nine escalated issues, including two escalations to a human. Coverage was reported equivalent to PostgreSQL.
More details
Remaining steps
- Implement and return work through review and escalation.
- Merge reviewed pull requests and compare dialect test coverage.
About this account
The pipeline planned and implemented Db2 support through specialized coding and review roles, filing further issues as defects and gaps appeared.
Platform: Blog. Attribution: https://www.cockroachlabs.com/blog/experiment-running-hospital-code/. Prompt: Suggested (written by Agents In Use, not the author's). Sourced by the Agents In Use pipeline.
Credit: rafiss on Blog
Source published
Added to Agents In Use
Source checked
Tools and topics
MOLT Sinai · GitHub · MOLT · ANTLR · Docker
Database migration · Db2 · Agent reviews
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.