setup-sensemaking-skills

Bootstrap skill that configures a repository to safely operate the sensemaking pipeline. It creates the necessary agent context and configuration artifacts.

Workflow

1. Explore: Analyze the repository for existing agent instructions (AGENTS.md, CLAUDE.md, .cursorrules) and documentation structures.
2. Identify Context: Locate the issue tracker, domain documentation, and existing skill ecosystems.
3. Interactive Decisions: Ask the user one decision at a time:
   • Target instruction file (e.g., AGENTS.md).
   • Default sensemaking path (fast_path vs full_fog_path).
   • Permitted execution modes (safety gates).
   • Downstream ecosystem mappings.
4. Draft Changes: Show the exact file modifications and new files to be created.
5. Execute: Write files only after explicit user approval.

Artifacts Managed

• AGENTS.md / CLAUDE.md (Sensemaking Block)
• docs/agents/sensemaking.md
• docs/agents/workflow-modes.md
• docs/agents/artifact-contracts.md
• docs/agents/downstream-skills.md

Boundary Rules

• Non-Diagnostic: This skill does not run the sensemaking pipeline or analyze project fog. It only handles configuration.
• Approval Mandatory: Never write or modify files without showing the diff and receiving approval.
• One at a Time: Ask interactive questions sequentially, not as a bulk list.

References

• Agent Block Template
• Sensemaking Config Template
• Workflow Modes Template