Context Packs

Context Packs are saved bundles of vault pages that an AI client can pull as setup context for recurring work. Instead of explaining your project, your team, or your operating context every conversation, you build a pack once and your AI loads it on demand.

Why packs exist

Most AI conversations start with re-explaining context. "I'm working on X. The constraints are Y. Here's the relevant background." That's wasted effort if you've already written it down somewhere in your vault.

Context Packs let you take that effort once and turn it into a named, reusable bundle. Your AI can then ask for the onboarding-redesign pack or the company-glossary pack and get every relevant page in one tool call.

Anatomy of a pack

A Context Pack has:

  • Name — a short identifier the AI uses to load it (e.g. onboarding-redesign)
  • Description — what the pack is for
  • Pages — the vault paths included in the pack

Pages in a pack are the actual current versions of those files — not snapshots. If you edit a page that's part of a pack, the next time the pack loads it includes the latest version.

Creating a pack

The web UI for managing Context Packs is part of the Agents workspace. From the macOS app's Agents view, switch to the Context Packs subview.

You can also create and manage packs via the API or via your AI:

  • mindwiki_list_context_packs — list all packs
  • mindwiki_get_context_pack — load a named pack with all its pages

Using a pack with your AI

When your AI client supports MCP, you can prompt it with something like:

Load the onboarding-redesign context pack from MindWiki and 
help me think through the next milestone.

The AI calls mindwiki_get_context_pack, receives the bundled pages, and uses them as immediate context for the conversation. No copy-paste required.

You can also have your AI plan its own context retrieval with the related mindwiki_get_relevant_context tool, which takes a goal and (optionally) constraints and returns pages it thinks are most relevant. Use Context Packs when you know the right pages; use mindwiki_get_relevant_context when you want the AI to pick.

Good pack patterns

A few uses where packs pay off:

  • Project briefs. A pack containing the project README, current status doc, key decisions, and stakeholder map. Anytime you start an AI session about that project, load the pack.
  • Operating manuals. A pack with your style guide, voice doc, content guidelines. Useful when you want AI writing to match your standards.
  • People context. A pack of people pages for a team, so an AI helping you draft a message has the relevant relationship context.
  • Recurring rituals. A pack for weekly reviews, monthly retros, quarterly planning — the same set of pages you always reference for that ritual.

Pack vs. just searching

Packs are not a replacement for mindwiki_search or mindwiki_get_relevant_context — they're a complement. Use search when the question is open-ended. Use packs when the right context is known and you don't want the AI to gamble on retrieval.

Where to go next

  • MCP Tools — mindwiki_list_context_packs, mindwiki_get_context_pack, mindwiki_get_relevant_context
  • Agents Overview — where Context Packs fit in the Agents workspace
  • Sessions & Activity — see when AI clients load your packs