narrative-writer
Skill0 starsChanged 45 days agoMITView on GitHub
Draft a written narrative of events for a custody evaluator, therapist, guardian ad litem, or attorney from the court timeline. Use when the user says "write my narrative", "draft the history for the evaluator", "statement of facts", "summarize what happened between X and Y", or needs a written account of a period.
In this repository10 other files it gives its agents.
- adhd-prioritiesSkill
- attorney-packetSkill
- co-parent-replySkill
- incident-logSkill
- proof-checkSkill
- repo-structure-auditSkill
- review-criticallySkill
- source-mapSkill
- timeline-builderSkill
- visual-explainerSkill
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 3 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- could be automatic: a setting could enforce it
- Very specific: line 23
Read
court/timeline.md: If it's missing or out of date, suggest running Timeline Builder first. - Distinctive: line 33
Source index: every cited ID and source, with date, what it is, and where it lives.
- Very specific: line 40
If a project tracker exists (PROJECT.md with Sections, Sources needed, Sources linked, Open questions, Last session), open from it when …
- Distinctive: line 46
No sarcasm, no adjectives that argue ("outrageous," "obviously").
- Distinctive: line 48
Never invent, merge, or embellish events.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Narrative Writer
Court folder: court/ below means whichever folder the user chose, recorded once as court-folder: in that folder's or the workspace's agent instructions. If unknown, ask once and record it. Never create a second one.
Turns the timeline into a calm, readable account that an evaluator or attorney can trust. Credibility comes from being factual, child-focused, and sourced.
Organized draft for your attorney to review, not legal advice.
Before writing, ask (one message)
- Who is it for? (custody evaluator, guardian ad litem, therapist, attorney)
- What period? (all, or dates)
- Any page or word limit, or required format?
- Anything they specifically asked about?
Steps
- Read
court/timeline.md. If it's missing or out of date, suggest running Timeline Builder first. - Select events: all 🔴 and the 🟡 entries that show a pattern. Leave ⚪ out unless needed for context.
- Write in chronological order, one paragraph per event or tightly linked cluster. Each paragraph ends with its sources in brackets, e.g.
[T-014; EVT-031; text 2023-05-04]. - Show patterns plainly: "This was the fourth late exchange in six weeks [T-020, T-022, T-025, T-027]."
- Keep the children at the center: what they experienced, how they responded, what the user did to protect their routine and wellbeing.
- Mark memory-only content: "To the best of my recollection…" for anything tagged
recollection. - List open questions at the end instead of filling gaps: "I don't have records for spring 2021. Do you have texts, emails, or someone who remembers?"
- Save to
court/narratives/YYYY-MM-DD-[audience]-[period].mdand show the user a short summary. - Run Proof Check on the saved draft. Don't call it ready until it says Ready for attorney review. Until then, call it a draft and list what's open.
Offer export to Word or PDF, using whichever document tool the app has. If there is none, say so. Put the disclaimer at the top, then add two appendices:
- Source index: every cited ID and source, with date, what it is, and where it lives.
- Timeline appendix: the cited rows from
court/timeline.md, unchanged.
The Markdown stays the master. Make changes there and export again. If the user wants an export before Proof Check passes, mark its first page
Not yet proof-checked.
Project tracker
If a project tracker exists (PROJECT.md with Sections, Sources needed, Sources linked, Open questions, Last session), open from it when resuming and update it at the end of each session. Don't create one unasked.
Tone rules
- First person, plain language, short sentences.
- Describe behavior, never diagnose or label ("the other parent arrived 90 minutes late," not "the other parent is a narcissist").
- No sarcasm, no adjectives that argue ("outrageous," "obviously").
- Acknowledge the user's own actions honestly; evaluators notice one-sided accounts.
- Never invent, merge, or embellish events. Never change what a quote says.
Rules
- Privacy. Files stay in the user's own folder and their own private backup, and nothing is sent to anyone else. Reading a file sends its contents to the AI provider, so check the app's privacy settings and ask the attorney.
- No unattended tidying. Automated tidy or lint jobs may report, but must not merge, move, rename, or rewrite anything in the court folder. State it in the folder's agent instructions, under
court-folder::automated-jobs: report only; never merge, move, rename, or rewrite anything here.
Output shape
> Organized draft for your attorney to review, not legal advice.
# [Narrative title] — [period]
Prepared for: [audience] · Draft date: YYYY-MM-DD · Based on timeline vN
## Summary (3–5 sentences)
## Background (brief)
## Events
[chronological paragraphs with sources]
## Patterns
## Impact on the children
## What I'm asking for / concerns going forward (only if the user wants this section)
## Open questions and gapsSource: structure follows the statement-of-facts variant of Anthropic's chronology skill (anthropics/claude-for-legal, Apache License 2.0): 🔴 and relevant 🟡 events, chronological prose, each paragraph cited. Child-impact focus and attribution standards from Case Forge (CaseForgeHq/family-court-strategist, MIT License).
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

