---
title: mcp tagged articles
canonical_url: https://www.phasoric.com/blog/tags/mcp
markdown_url: https://www.phasoric.com/blog/tags/mcp.md
---
# mcp tagged articles

- [How to Resume an Interrupted AI Agent Task Without Losing the Audit Trail](https://phasoric.com/blog/resume-interrupted-ai-agent-task-handoff-checklist), Recover an interrupted AI agent task safely with a portable state ledger, verified execution evidence, idempotent retries, and a 35-minute handoff drill.
- [Local vs Hosted AI Memory: Choose the Right Vault Setup for MCP](https://phasoric.com/blog/local-vs-hosted-ai-memory-vault-mcp-guide), Decide between local-only, provider-backed, and hosted AI memory. Compare MCP access, storage boundaries, permissions, and a 25-minute verification workflow.
- [How to Resolve Conflicting AI Knowledge Without Erasing History](https://phasoric.com/blog/resolve-conflicting-ai-knowledge-sources), Two AI knowledge sources disagree. Classify the conflict, compare effective dates and evidence, and resolve it without deleting valid history. Includes a 30-minute review workflow.
- [AI Memory Retention Policy: What to Keep, Review, and Delete](https://phasoric.com/blog/ai-memory-retention-policy-checklist), Build a practical AI memory retention policy with record classes, review dates, superseded decisions, local Trash controls, and a safe deletion checklist.
- [How to Test AI Memory Accuracy: A 12-Question Evaluation Checklist](https://phasoric.com/blog/test-ai-memory-accuracy-evaluation-checklist), Build a repeatable 12-question AI memory test for source recall, current decisions, access boundaries, citations, and honest abstention—without trusting a polished summary.
- [How to Give an AI Agent Write Access to Notes Without Losing Control](https://phasoric.com/blog/safe-mcp-ai-agent-note-write-access), Safely connect AI agents to editable notes: compare MCP read/write permissions, preview approvals, expected hashes, idempotent retries, and deletion safeguards.
- [AI Knowledge Backup and Restore: A 30-Minute Vault Recovery Drill](https://phasoric.com/blog/ai-knowledge-backup-restore-recovery-drill), An AI knowledge backup is only useful if it restores. Run a 30-minute, non-destructive vault recovery drill covering Markdown notes, attachments, links, decisions, and MCP permissions.
- [How to Share Project Context Between ChatGPT and Claude Without Copying Chats](https://phasoric.com/blog/share-project-context-between-chatgpt-and-claude), Move project context between ChatGPT and Claude without copying chat transcripts. Use reviewed Markdown, authorized MCP retrieval, and a five-case handoff test.
- [AI Agent Using Outdated Context? Trace the Wrong Answer to Its Source](https://phasoric.com/blog/ai-agent-outdated-context-source-audit), An AI agent can cite a real note and still follow an obsolete decision. Audit source permissions, evidence, revisions, and temporal validity before correcting the answer.
- [How to Prevent AI Agents From Mixing Client Data](https://phasoric.com/blog/prevent-ai-agents-from-mixing-client-data), Prevent cross-client AI data leaks with scoped retrieval, audience-aware permissions, safe memory review, revocation checks, and practical isolation tests.
- [MCP Connected but Can't Find Notes? A Troubleshooting Guide](https://phasoric.com/blog/mcp-connected-but-no-notes-troubleshooting), Your AI client connects through MCP but cannot find known notes. Diagnose authentication, mappings, Context Scope, local-only data, provider permissions and search.
- [What Is MCP? Connecting AI Assistants to Your Knowledge Base](https://phasoric.com/blog/what-is-mcp-for-knowledge-bases), Model Context Protocol (MCP) is a standardized way for AI clients to use tools and access resources exposed by compatible servers. For personal and team knowledge, its practical value is the ability to ask different assistants for authorized informat
- [Stop Starting Over With Your AI](https://phasoric.com/blog/stop-starting-over-with-your-ai), AI models keep getting smarter, but every new session can still feel like starting over. Durable project context may be the missing layer between powerful AI and useful long-term collaboration.
