Run any Skill in Manus with one click

Get Started

prune-context

Stars10

Forks3

UpdatedJune 19, 2026 at 08:50

Claude Code context pruner: slash command to summarize session and reset token bloat

Installation

Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.

Run Skill in Manus

Source

UitbreidenOS

UitbreidenOS/Claudient

View GitHub Repository View Creator Repositories

Download

Run Skill in Manus

Related occupationsSOC

Based on SOC occupation classification

Software DevelopersComputer and Mathematical Occupations·SOC 15-1252

SKILL.md

readonly

name	prune-context
description	Claude Code context pruner: slash command to summarize session and reset token bloat

Context Pruner

When to activate

Activate when the user runs the /prune-context command, or when session token consumption is high, or when the context window is getting bloated in a long session.

When NOT to use

Do not use at the beginning of a session, or for simple queries that do not require state persistence or context reduction.

Instructions

Pause current tasks and review the conversation history of the active session.
Summarize the session into a highly condensed, token-efficient state:
- Active Objective: The current high-level goal you are working on.
- Completed in Session: A concise bulleted list of completed changes, tests, and files.
- Key Architectural Context: Crucial project details, APIs, or files discovered during the session.
- Next Steps: Explicit checklist of what needs to be done next.
Write this summary to .claude/pruned_context.md.
Output the summary in the chat so the user can see it.
Instruct the user to run the /clear command to reset the context window and wipe the token bloat.
Provide the command or text they should input after clearing to resume: "Please run /clear now to reset memory. Once cleared, prompt me with: 'Resume from .claude/pruned_context.md' to restore this state."

Example

User: /prune-context Claude: Starting context pruning to reduce token bloat. [Claude writes summary to .claude/pruned_context.md]

Current Session Snapshot

Active Objective: Restructure README documentation and add fail-fast enforcer rules.
Completed: Created rules/common/fail-fast.md and hooks/post-tool-use/fail-fast.sh.
Key Context: The compiler is run via npm test and require specific git authoring config.
Next Steps: Implement Plan-First pre-tool-use hook.

State saved to .claude/pruned_context.md. Please run /clear now to reset memory and clear token bloat. Once cleared, prompt me with: Resume from .claude/pruned_context.md to restore this state.

More from this repository

same repository

agent-teams

UitbreidenOS/Claudient

Orchestrate multi-agent teams in Claude Code — set up coordinated sessions with task delegation, inter-agent communication, and parallel execution

2026-06-2310

ultraplan

UitbreidenOS/Claudient

Leverage Claude Code's ultra-deep planning mode for complex architecture decisions, multi-file refactors, and system design

2026-06-2310

ultrareview

UitbreidenOS/Claudient

Deep code review using Claude Code's thorough analysis mode — security, performance, correctness, and maintainability

2026-06-2310

swarm-sandbox

UitbreidenOS/Claudient

Safe isolated testing environment for multi-agent swarm topologies before production deployment

2026-06-2210

auto-summarizer

UitbreidenOS/Claudient

Automatically summarizes the current session context to prevent token window overflow.

2026-06-2210

campaign-tracker

UitbreidenOS/Claudient

Produces a campaign performance summary from session-log.md entries. Reports open rates, reply rates, and meeting conversion rates per sequence. Flags top performers (reply rate >10%) and underperformers (reply rate <3%). Recommends one concrete change per underperforming sequence.

2026-06-1710

Context Pruner

When to activate

Activate when the user runs the /prune-context command, or when session token consumption is high, or when the context window is getting bloated in a long session.

When NOT to use

Do not use at the beginning of a session, or for simple queries that do not require state persistence or context reduction.

Instructions

Pause current tasks and review the conversation history of the active session.

Summarize the session into a highly condensed, token-efficient state:

Active Objective: The current high-level goal you are working on.
Completed in Session: A concise bulleted list of completed changes, tests, and files.
Key Architectural Context: Crucial project details, APIs, or files discovered during the session.
Next Steps: Explicit checklist of what needs to be done next.

Write this summary to .claude/pruned_context.md.

Output the summary in the chat so the user can see it.

Instruct the user to run the /clear command to reset the context window and wipe the token bloat.

Provide the command or text they should input after clearing to resume: "Please run /clear now to reset memory. Once cleared, prompt me with: 'Resume from .claude/pruned_context.md' to restore this state."

Example

User: /prune-context Claude: Starting context pruning to reduce token bloat. [Claude writes summary to .claude/pruned_context.md]

Current Session Snapshot

Active Objective: Restructure README documentation and add fail-fast enforcer rules.

Completed: Created rules/common/fail-fast.md and hooks/post-tool-use/fail-fast.sh.

Key Context: The compiler is run via npm test and require specific git authoring config.

Next Steps: Implement Plan-First pre-tool-use hook.

State saved to .claude/pruned_context.md. Please run /clear now to reset memory and clear token bloat. Once cleared, prompt me with: Resume from .claude/pruned_context.md to restore this state.