compound

Parse the task file and append session learnings under future-agent-notes.

3|1|Updated Jul 2, 2025

Search Tags:#memory #learnings #AGENTS.md #knowledge-base #compound #task-files #AI sessions

Ready to run

npx skills add https://github.com/benredmond/apex --skill compound-benredmond

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: compound
Source: https://github.com/benredmond/apex/tree/main/skills/compound
Command: npx skills add https://github.com/benredmond/apex --skill compound-benredmond

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Capture learnings (problems, decisions, gotchas) from AI sessions to improve future agents by preserving context and insights for reuse.

Core Features & Use Cases

Capture problems encountered, decisions made, and gotchas discovered during a session and write them to the task file for future retrieval.
Scan existing task notes to surface related learnings and avoid duplication.
Promote critical learnings to AGENTS.md for always-loaded context in future runs.
Use case: after a debugging or design-session, capture what happened, why decisions were made, and any surprises to guide future work.

Quick Start

Invoke the skill after completing a session, using a task identifier, for example: /apex:compound TASK-IDENTIFIER

Frequently Asked Questions about compound

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

FAQPage Schema

How do I capture learnings from AI sessions for future context?▼

To capture learnings from AI sessions, you parse the current task file and append detailed notes under a future-agent-notes section. This preserves problems, decisions, and gotchas to build a lasting knowledge base for future agents.

What's the best way to preserve architectural decisions in a knowledge base?▼

The best way to preserve architectural decisions is to scan existing task notes to surface related learnings and avoid duplication. Optionally, you can promote critical decisions to an AGENTS.md file for always-loaded context in future runs.

When do I need to write session notes to a task file?▼

You need to write session notes to a task file after completing a debugging session, making an architectural decision, or finishing any significant problem-solving. This captures what happened, why decisions were made, and any surprises to guide future work.

Does AGENTS.md work with task files to load context for AI agents?▼

Yes, AGENTS.md works with task files by providing always-loaded context for future AI agents. Critical learnings can be promoted from task files into AGENTS.md, while related tasks are linked back to ensure continuous knowledge retrieval.

Can I avoid duplicating knowledge when appending future-agent-notes?▼

Yes, you can avoid duplicating knowledge when appending future-agent-notes by scanning existing task notes first. The process surfaces related learnings already captured in the task file, ensuring new notes add distinct context without redundancy.