Project Workflows

Turn saved knowledge into better research, clearer plans, and work another agent can continue. MindWiki is most useful when the brief, evidence, decisions, and next action stay together in a vault you and your agents can retrieve.

You can use these workflows with one agent or divide them among several. Start with a single project and add structure when it saves effort.

Research to a useful brief

Outcome: source notes and a clear brief that remain available to future work.

1. Save the question you are investigating.
2. Capture source notes with links, dates, and the reason each source matters.
3. Ask an agent to retrieve the sources and compare the evidence.
4. Save a brief separating findings, interpretations, and unanswered questions.
5. Link the brief from the project hub.

Read the MindWiki sources for our onboarding research. Create a brief that separates supported findings, competing explanations, and open questions. Cite the source pages and tell me where you save the draft.

Keep original attachments beside companion summaries when useful. A readable summary gives retrieval something precise to find, while the original file preserves the supporting material.

Plan to implementation handoff

Outcome: a coding agent starts from current requirements and leaves a usable record of the result.

Keep the project goal, requirements, chosen approach, and constraints in separate clear sections or linked pages. Ask the coding agent to read them before changing anything.

Retrieve the current Website Launch brief and decisions from MindWiki. Summarize the requirements and constraints before implementing. After the work, save what changed, what was tested, any release status, and the next action.

Distinguish proposed, implemented, tested, and released work in the handoff. This prevents the next session from treating an unfinished draft or unverified claim as a completed outcome.

Meeting notes to durable decisions

Outcome: a conversation becomes an actionable project record.

Capture the meeting notes quickly. Then separate the decision, reasoning, owners, and next steps. Link the decision to the relevant project and any evidence discussed.

Read this meeting note in MindWiki. Draft a concise decision record with the choice, alternatives, reasons, unresolved questions, and next actions. Preserve uncertainty where the note does not contain an answer.

A checkbox records a task in a note; it does not automatically assign it in an external task manager. Dates and owners remain useful context even when no automation acts on them.

Research to writing

Outcome: a draft uses your evidence, audience context, and preferred voice.

Maintain a small writing pack containing the audience, voice guide, current brief, and key source notes. Ask the writer to load that context before drafting. Save the draft as its own page so you can compare it with the source material.

Prepare MindWiki context for writing our launch announcement. Use the audience brief, voice guide, and current product decisions. Draft the announcement, identify unsupported claims, and save it separately from the source notes.

The pack improves the starting material; you still review the draft for correctness, fit, and voice.

One project across several agents

Outcome: each agent contributes its strengths without you recopying every background document.

A researcher saves evidence. A planning agent reads it and writes requirements. A coding agent retrieves the requirements. A writing agent uses the same decisions to prepare release copy.

Connect each client to your existing account. Give it a specific role, a source hub or pack, and a separate output page. Ask it to save a handoff when done. This makes parallel work easier to reconcile and reduces two agents editing the same document at once.

A recurring project review

Outcome: a brief you can trust at the start of the week.

Prepare my MindWiki context for the weekly Website Launch review. Read current status, decisions, and recent handoffs. Summarize progress, blockers, decisions needed, and the next three actions. Cite the pages used and propose corrections separately.

Save the review and update the current brief deliberately. Use a Context Pack if the same sources repeatedly support the review. Scheduled automations can supply maintenance suggestions and a monthly summary, while your interactive agent handles the task you request.

Build a small project home

projects/website-launch/
  overview.md
  brief.md
  decisions.md
  research-summary.md
  handoff.md

These names are examples. The overview can link the important pages with a sentence explaining each. Begin with fewer pages if the project is small; add separate source notes and attachments as the work develops.

A useful workflow leaves three things behind: a result you can inspect, evidence you can follow, and a clear next action. That is what lets your next session build on the last one.

Next: Shared Agent Memory, Context Packs, and Vault Structure.