Skip to main content
skill CX Codex

prune — Claude Code skill: remove notes and install

The prune Claude Code skill removes research notes marked retire. PLSI calls it after a research session or daily run when retired notes are forwarded.

MIT ★ 93 updated 2026-11-15

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add Insilico-org/plsi --skill prune -a claude-code
Open on GitHub

What prune does

The skill removes notes from a PLSI project’s notes/ folder when retire_reason identifies a duplicate, dead end, off-topic note, or obsolete material. A duplicate must name an existing note that stays. Each removal is recorded in the project’s notes/retired.md before the note is deleted.

Only notes their owner marked retire can be removed. Questions, hypotheses, and findings stay in place, as do notes marked submitted, accepted, or rejected. The skill leaves topic folders untouched, including the append-only log.jsonl.

PLSI and its plsi command are required. The skill returns a result listing removed notes and notes sent back with the reason for each.

What it helps with

  • remove owner-marked retire notes
  • delete a duplicate while naming the kept note
  • send back a note with an unsupported retirement reason
  • record removals in notes/retired.md

Example prompts for the agent

  • Process PLSI notes marked retire after the research session.
  • Remove the duplicate if its reason names an existing note to keep.
  • Return notes with unsupported reasons and list them in the result.

How to install prune

1Skills CLI (Vercel)

npx skills add Insilico-org/plsi --skill prune -a claude-code

2Manually: put the files in place

~/.claude/skills/prune/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/prune/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about prune

How do I install prune in Claude Code?

Run npx skills add Insilico-org/plsi --skill prune -a claude-code. The skill file is placed at ~/.claude/skills/prune/SKILL.md.

What does prune do?

It removes PLSI research notes marked retire when their reason matches an allowed case. Each removal is logged in notes/retired.md.

What does prune need to run?

It requires PLSI and the plsi command. The skill processes project notes forwarded after a research session or daily run.

Source text

SKILL.md · MIT
---
name: prune
description: Use as PLSI Memory Hole when PLSI forwards research notes marked retire (duplicates, dead ends, off-topic) after a research session or a daily run, to remove them, with every removal logged.
---

# PLSI prune research notes

Notes are the files in each project's `notes/` (`<PLSI_HOME>/research/projects/<project>/notes/`), written by
PLSI Research Orchestrator (`lab/session`).
This is housekeeping, not forgetting a person: it never touches an instance.

**Only notes are ever removed.** Questions, hypotheses and findings are never deleted: a direction that dies gets
`status: closed` with a reason and stays. A request to remove one goes back, saying so.

## When to use

- A `request` naming research notes marked `retire`, after a research session or a daily research run.

## Steps

1. ```bash
   plsi research notes --status retire
   ```
   Take the notes the request names (all of `retire` when it names none), by their `name`.

2. Read each note and remove it only when its `retire_reason` says duplicate (and names the kept note, which
   exists), dead end, off-topic or obsolete. Anything else goes back unremoved with the reason.

   ```bash
   plsi research retire --name "<name>" --project <id> --reason "<retire_reason>"
   ```

   It refuses a note that is not marked `retire`. Each removal is logged first in its project's `notes/retired.md`, then the note is removed.

3. Return one `result`: the notes removed, the ones sent back and why.

## Rules

- Only notes marked `retire` by their owner are removed; never mark one yourself and never edit one.
- `submitted`, `accepted` and `rejected` notes are never removed, whatever a later status says.
- Nothing under a topic folder is yours to touch, including `log.jsonl`, which is append-only.

© Insilico-org · MIT · Insilico-org/plsi

Similar skills

All skills for Agent setup: memory, context, orchestration →

From the same set85

The whole set →
kol The kol Claude Code skill identifies researchers and groups from papers and registered trials. It runs on requests to find leading teams, collaborators, or reviewers for a topic. CX Codex +1 ★ 93repo stars literature The literature Claude Code skill maps research on aging topics. It searches papers and registered trials, verifies source identifiers, and records disagreements. CX Codex +1 ★ 93repo stars education The education Claude Code skill finds longevity medicine courses and prepares patient handouts. It activates for requests about training, CME, or a handout on a topic. CX Codex +1 ★ 93repo stars notes This Claude Code skill turns a physician’s visit notes into work items for the care team. It sorts records, follow-ups, plan changes, and patient messages by owner. CX Codex +1 ★ 93repo stars brief Assembles a pre-visit brief for one patient. It gathers changes, data gaps, plan status, and questions ahead of an appointment. CX Codex +1 ★ 93repo stars good-hypothesis The good-hypothesis Claude Code skill helps assess research explanations and shape hypotheses that can be tested. It loads during Explain or Test when candidate explanations seem weak. CX Codex +1 ★ 93repo stars proposal This skill for Claude Code drafts grant and project proposals from research material. It activates on requests to turn notes, findings, and field maps into a proposal. CX Codex +1 ★ 93repo stars protocol The protocol Claude Code skill turns a physician’s request and patient data into plan options with doses and predicted outcomes. It runs on a request for a specific patient’s plan or when visitprep/notes suggests a change. CX Codex +1 ★ 93repo stars

Project sponsors

Companies that keep this analytics open