All agent instructions are maintained in CLAUDE.md. Refer to that file for documentation references, external tool usage (dotconfig, rundbat, clasi), testing requirements, and the SE process.
This project uses the CLASI (Claude Agent Skills Instructions) software engineering process, managed via an MCP server.
The SE process is the default. When asked to build a feature, fix a bug, or make any code change, follow this process unless the stakeholder explicitly says "out of process" or "direct change".
Work flows through four stages organized into sprints:
- Requirements — Elicit requirements, produce overview and use cases
- Architecture — Produce technical plan
- Ticketing — Break plan into actionable tickets
- Implementation — Execute tickets
Use /se or call get_se_overview() for full process details and MCP
tool reference.
When the stakeholder corrects your behavior or expresses frustration ("that's wrong", "why did you do X?", "I told you to..."):
- Acknowledge the correction immediately.
- Run
get_skill_definition("self-reflect")to produce a structured reflection indocs/plans/reflections/. - Continue with the corrected approach.
Do NOT trigger on simple clarifications, new instructions, or questions about your reasoning.