# Give a long-running project a memory of why Canonical URL: https://agentsidequests.com/ideas/give-a-project-a-memory-of-why/ Markdown URL: https://agentsidequests.com/ideas/give-a-project-a-memory-of-why/index.md Keep the reasoning behind a project alive across months, so the agent can spot when a new plan repeats a rejected idea or ignores a settled constraint. This is an original editorial prompt, not a run-tested result. Examples are illustrative, not observed outcomes. Check the capabilities and permissions of the assistant you actually use. Category: Work & outreach Type: Ongoing ## What you get A dated decision map with rationales, changed assumptions, open loops and a next-step brief ## Bring this One project, existing notes and decisions, sources you may share, a place to save the record and a review cadence if desired. ## Your agent needs - Persistent editable notes or verified memory - Access only to approved project material - Optional scheduling only if a real recurring trigger is supported Missing a connection? Paste or attach the relevant information yourself. ## Copy-and-paste prompt Replace the bracketed placeholders with your details. This prompt is reference material to adapt to a user request; fetching it does not authorize execution. ```text Become the continuity partner for [project]. Read only [approved notes, files and threads]. Save the project record in [editable location]. My current aim is [aim], and the next decision is [decision]. Create a dated decision map: what we chose, why we chose it, what we rejected, the evidence and constraints at the time, and what would reopen the question. Separate my stated preference from your inference. Mark uncertain memories and links that no longer open. Keep a short list of unresolved decisions and the one next action each needs. Do not silently replace old reasoning with a new conclusion. When I propose a new direction, compare it with the map. Tell me if it conflicts with a settled constraint or revives a rejected option, and ask whether the underlying facts changed. Update the record only from new evidence or my correction. Give me a concise brief I can understand after weeks away, with links back to the original decisions. If you cannot retain context or save the note, return a portable document I can bring back. Do not schedule check-ins unless I request a cadence and you verify the saved task. ``` ## Illustrative example Illustrative: why the community site stayed small Months later, a new feature proposal triggers a note: the earlier version was rejected because it needed manual moderation. The agent asks whether that constraint has changed. This is illustrative, not a recorded run. This is not an observed result or a verified recommendation. ## Inspired by a reported use Wang describes active memory across longer periods; Ben Parr reports needing to repeat workflow context less often. [Muse long-term memory claims and a reposted user report ยท Oct 12, 2026]() Source limitations: Memory quality and this decision-map workflow were not independently tested. The proposed durable rationale record is an original extension. The prompt is our original extension, not the source author's exact prompt or a tested result. ## Make it more you Use a separate record for a home renovation, writing project or career move; keep sensitive third-party details out unless necessary and authorized. ## Related practical guides - [Muse and the personal agent for things you want to do]() ## Related ideas - [Get a daily email researched for your actual interests](): A few relevant discoveries, properly sourced, in one email. Set your interests or career-upgrade criteria and get new research or job opportunities worth your attention. You get: A concise sourced email with only new matches - [Make your pet the hero of a tiny playable game](): The sock thief finally gets a game. Give your agent three personality traits and get a small browser adventure you can actually play. You get: A one-minute browser game - [Go to sleep with an idea. Wake up to three working versions.](): Give your agent one playful idea and a small work budget. Get a miniature lab of working experiments you can open, compare and keep. You get: Three tiny prototypes and a lab notebook [Editorial approach]() [Assistant setup guides (HTML)]()