$ squidy --docs
Framework 🦑 Squidy
A set of 10 Markdown documents that give your AI agent permanent context, rules and traceability — in any session, with any LLM.
How it works
# The agent runs this ritual before any action:
1. Read readme-agent.md → confirm initialization
2. Read doc/AGENT.md → agent instructions
3. Read doc/constitution.md → load rules
4. Read doc/oracle.md → load ADRs
5. Check doc/emergency.md → verify blockers
6. Check doc/kanban.md → pick a task
7. Read doc/session-context.md → current state
✓ Agent operational with full context
The 10 documents
readme-agent.mdEntry point. Instructs the agent to execute the initialization ritual before any action.
doc/AGENT.mdSpecific instructions for Claude, Cursor and other agents. Loaded automatically in AGENT sessions.
doc/constitution.mdPrinciples, prohibitions and Definition of Done. The agent consults it before each decision.
doc/oracle.mdArchitecture Decision Records. Documents the reasoning behind each relevant technical decision.
doc/kanban.mdTask board with sequential IDs, acceptance criteria and progress traceability.
doc/emergency.mdCritical blocker log. Prevents the agent from moving forward on known issues.
doc/session-context.mdCurrent project state: what was done, what remains and what the next step is.
doc/policies.mdCode conventions, naming patterns and style standards to follow.
doc/diary-index.mdConsolidated index of all development diary entries.
diary/YYYY-MM.mdChronological record of decisions and actions. The agent writes what it did and why.
Available commands
squidy initInitialize governance in the project (interactive interview)
squidy init --lang en-USGenerate documents in English
squidy init --dry-runPreview files without creating them
squidy auditAudit consistency of existing documents
squidy audit --fixAutomatically fix detected inconsistencies
squidy statusDisplay current project and document status
squidy doctorFull diagnostic of installation and configuration