One-click install
npx skills add https://github.com/harnessprotocol/harness-kit --skill capture-harnessprotocol
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: capture
Source: https://github.com/harnessprotocol/harness-kit/tree/main/plugins/capture/skills/capture
Command: npx skills add https://github.com/harnessprotocol/harness-kit --skill capture-harnessprotocol

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

It prevents important decisions and technical details from being lost after a conversation ends by staging them in a consistent, append-only file for downstream reflection and knowledge-graph updates.

Core Features & Use Cases

  • Manual fact capture: Create timestamped bullet entries from the current conversation via /capture, including a confirmation of what was staged.
  • Auto-extraction: Capture 3–8 most important facts when called with no arguments.
  • Targeted filtering: Stage only decisions or only technical details using simple keywords.
  • Compatibility with session-end automation: Works alongside a Stop hook (session-end.sh) that performs reflection at session end while deduplicating via manual markers.

Quick Start

Use the capture skill to stage decisions from your current conversation by writing: /capture decisions.

Frequently Asked Questions about capture

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I save key conversation decisions and technical notes before a session ends?

You can capture conversation decisions and technical notes by appending timestamped bullet entries to a staging file, preserving important facts for later reflection before the session ends.

How does append-only logging work for capturing conversation facts?

Append-only logging for conversation facts works by resolving a staging path and adding new timestamped entries with a manual source marker, ensuring prior records remain unchanged for downstream knowledge-graph processing.

Can I extract only technical details or decisions from a coding or planning session?

Yes, you can extract only technical details or only decisions from a coding session by passing a targeted keyword to the capture command, which filters and stages the specific subset of facts you need.

What is the best way to auto-summarize important facts from an ongoing conversation?

The best way to auto-summarize important facts is to invoke the capture command with no arguments, which automatically extracts the 3–8 most important facts from the current conversation and stages them.

Does manual fact capture work with session-end reflection hooks?

Yes, manual fact capture works with session-end reflection hooks by using manual source markers to deduplicate entries, allowing automated end-of-session processing to run alongside on-demand staging without conflicts.

Where are staged conversation facts stored when capturing technical notes?

Staged conversation facts are stored in an append-only markdown file located at the project scripts directory or a fallback home directory path, ensuring consistent access for downstream knowledge-graph updates.