Personal Content DNA

Capture real decisions and feedback to develop an authentic writing voice

Turn real personal input from exploration into traceable materials, then practice writing with reference methods. Material capture and expression training can start together; a mature voice is not a prerequisite.

Choose a mode

• Capture: for session wrap-up, specified-record review, GitHub exploration, material confirmation or weekly reconciliation, read references/capture.md. Use assets/material-card.md.
• Writing practice: for topic selection, briefs, drafts, personalized revision or selection feedback, read references/writing.md. Use assets/style-rule.md.

A request can combine modes. Preserve the project's topic, research and review gates.

Library and permitted scope

Read project rules first. Choose the library path in this order: user-specified path, project AGENTS.md, then reference/personal-dna/ in the current project. Create an index and only the cards needed for actual work. Cross-project aggregation follows the user's specified source and destination.

Automatic skill selection only affects an active session. Continued capture relies on milestones, wrap-up, confirmation and editing feedback. This skill has no background monitor, scheduled collector or unrestricted cross-session access. Installation is not evidence of a running background process.

Capture authorization comes from the request or an explicit project candidate-capture rule. Default scope is the current visible conversation and specified files/public repositories. Other conversations, complete histories, private repositories and internal records require an explicit corresponding scope. Use supported public-network tools; retain GitHub versions or checked dates.

Personal input and evidence

Separate user quotations, AI proposals, user selections/rejections, AI execution and actual results. AI-assisted exploration can supply personal material when roles and judgment origins are clear. Repository ownership, successful commits, silence and acceptance of wording do not establish independent development or adoption of a new lasting opinion.

Save the event's problem, options, user tradeoff, action, result evidence, changed view and reader use. Ground aesthetic preferences in concrete version comparisons or explicit edits. Keep failures, tentative judgments and unknowns. Do not infer psychological traits, unexpressed motives or emotions.

Do not copy private/internal text or credentials into ordinary candidate cards or routine drafts. Safe excerpts within the permitted scope become candidates; otherwise record the gap. Do not pass restricted original records to a Writer.

States and confirmation

• Material: candidate, approved, needs_review, historical.
• Expression rule: trial, adopted, retired.
• Public use: pending, allowed, restricted.

New materials default to candidate + pending. If the user has already confirmed meaning and public scope, record the quotation/location and update without asking for the same permission again. Only approved + allowed material cards enter routine writing; expression methods may be practiced as trial + allowed. Synchronize and read back the index after changes. Public scope does not authorize publication.

Writing and learning

Use relevant allowed materials and current facts, then choose methods that fit the topic. x-writing-dna can supply mechanisms, genres, supporting works and limits. If unavailable, use supplied references or the user's practice goal without inventing author-research findings.

Honor the user's positioning and current post/long-form definitions. Choose one organizing principle; supporting methods serve it. Personalization must add concrete context, tradeoffs or judgment evidence. Without relevant lived experience, use public research or explicitly proposed judgments; never manufacture first-person experience. Place new opinion-claiming items in review materials, not the clean body.

Keep material cards and reference methods separate. Actual selections and edits support method adoption. Draft text, AI summaries, publication and interaction data do not prove real experience. One preference supports that selection; a stable rule needs cross-topic feedback or explicit adoption.

Save and report

Inspect the existing index and event before adding a card. Merge supplements with a change history; paraphrasing does not create a new event. Read the latest version before updating. If another writer has changed it, retain a separate new card rather than overwriting their work.

Save and read back cards, index and actual feedback. Report new/updated IDs, source scope, states, gaps and next action. No meaningful new material means no forced card. Distinguish format validation, actual writing adoption and distribution outcomes.