Claude Code plugins for epistemic dialogue. Transforms unknown unknowns into known unknowns. Benefits operations teams by identifying blind spots. Integrates with Claude for AI collaboration.
git clone https://github.com/jongwony/epistemic-protocols.gitEpistemic Protocols inserts structured checkpoints across planning, analysis, decision, execution, verification, and comprehension to surface misalignment between you and AI before it hardens into code or downstream work. The skill spans 14 protocols—including Aitesis for clarifying unstated needs, Prosoche for execution guardrails, Anamnesis for session recall, and Katalepsis for comprehension verification—each triggered by a slash command when specific collaboration problems arise. Rather than redoing entire implementations after discovering wrong direction, these protocols fix misalignment at the plan level where a single conversation turn can resolve what might otherwise cost hours of rework. Teams benefit by identifying blind spots early, keeping AI and human intent aligned throughout complex projects.
[{"step":"Install the epistemic-protocols plugin in Claude Code by running `claude plugins install epistemic-protocols` in your terminal.","tip":"Ensure you have the latest version of Claude Code (v1.4+) for full plugin compatibility."},{"step":"Prepare your prompt by specifying the team, project, and topic. Use the prompt template to structure the dialogue around uncovering assumptions.","tip":"For complex projects, break the topic into smaller sub-topics (e.g., 'user adoption,' 'technical feasibility') to focus the dialogue."},{"step":"Run the dialogue in a collaborative session with your team. Share the output table in real-time using Claude's multi-user mode or export it to Notion/Google Docs.","tip":"Encourage team members to challenge assumptions by asking 'What would convince you this assumption is wrong?' during the session."},{"step":"Schedule a follow-up review 1-2 weeks after the initial dialogue to revisit the 'Next Experiment' column and track progress.","tip":"Use the plugin's 'export' feature to generate a Markdown report for stakeholder updates."},{"step":"Iterate: Repeat the process for new blind spots or after major project milestones.","tip":"Combine with the 'risk-assessment' plugin to prioritize blind spots based on impact and likelihood."}]
Prevent wrong direction from hardening into code by catching misalignment at the plan level
Surface unstated AI assumptions and missing context before autonomous execution
Verify AI comprehension of major changes before approval or handoff
Recover and distill session context when transferring work across conversations
No install command available. Check the GitHub repository for manual installation instructions.
git clone https://github.com/jongwony/epistemic-protocolsCopy the install command above and run it in your terminal.
Launch Claude Code, Cursor, or your preferred AI coding agent.
Use the prompt template or examples below to test the skill.
Adapt the skill to your specific use case and workflow.
Use the epistemic-protocols plugin to facilitate a structured dialogue with [TEAM/STAKEHOLDER] about [PROJECT/INITIATIVE]. Focus on uncovering unknown unknowns by asking: 'What assumptions are we making about [TOPIC] that we haven't explicitly tested?' and 'What evidence would change our minds about [ASSUMPTION]?' Document the discussion in a table with columns: Assumption, Evidence For, Evidence Against, and Next Experiment. Share the output with the team and schedule a follow-up to review findings.
### Epistemic Dialogue: Launching New Customer Support Portal **Team:** Product, Engineering, Customer Success **Topic:** Assumptions about user adoption of the new support portal | Assumption | Evidence For | Evidence Against | Next Experiment | |------------|--------------|------------------|-----------------| | Users prefer self-service over live chat | 78% of survey respondents want 24/7 access | Only 30% of power users have tried the beta | Deploy A/B test: 50% of new users see portal homepage with self-service options first | | Support agents will adopt the new tool | Training sessions show 90% completion rate | Agents report 'too many clicks' in usability tests | Measure agent ticket resolution time pre/post-training | | Portal reduces ticket volume by 40% | Competitor portals show 35% reduction | Historical data shows seasonal spikes in tickets | Track ticket volume for 8 weeks post-launch | **Blind Spots Identified:** 1. **Integration with existing CRM:** No testing done on how portal data syncs with Salesforce. Next step: Schedule API integration review with Engineering. 2. **Mobile usability:** Only 15% of beta testers were mobile users. Next step: Conduct mobile-specific usability tests. **Follow-up Actions:** - Product team to prioritize mobile UX improvements by [DATE]. - Customer Success to survey 100 beta users on portal pain points. *Output generated using epistemic-protocols plugin in Claude Code.*
Your all-in-one knowledge management tool.
AI assistant built for thoughtful, nuanced conversation
Get more done every day with Microsoft Teams – powered by AI
Automate your spreadsheet tasks with AI power
Agentic AI Workflow platform
Connected workspace for docs, wikis, and projects
Take a free 3-minute scan and get personalized AI skill recommendations.
Take free scan