Agent Memory Discipline
Teaches an agent when to recall from long-term memory before acting and when to save decisions afterward.
- Repository
- sickn33/agentic-awesome-skills
- Stars
- 49,023
- Licence
- MIT
- Files
- 2 files
What it does
Connecting a memory tool does not make an AI agent use it well. This skill gives the agent standing rules: read relevant memory before starting project work, choosing tools or writing conventions, and save decisions, user corrections and failures afterward as short single-fact notes. It also tells the agent to keep old notes marked as superseded rather than deleted, and to show conflicting memories to the user instead of silently picking one. It works with a plain Markdown memory folder or any memory server the agent can call.
Great for
- Developers using a memory tool or MCP memory server whose agent keeps forgetting preferences or past decisions
- Teams setting up persistent project memory who need consistent rules for reading and writing it
- Users who repeatedly correct the same assistant habits across sessions
Skip it if
- Simple one-off questions, since the skill itself says recall adds cost and should be skipped for self-contained tasks
- Setups where memory must not be stored at all, because the skill writes notes after decisions and corrections
- Anyone expecting a ready-made memory product: it is a set of rules and does not provide storage itself
Try asking
- “Remember that this project uses pnpm, not npm, after I correct you”
- “What did we decide about the billing module last time? Check memory before changing it”
- “Save this for next time: the dev server needs Redis running first”
- “Before writing tests here, recall our testing conventions and tell me what you are relying on”
Install in 3 steps
- 1
Download the skill
The ZIP contains the 2 original files, the MIT licence and an attribution note.
agent-memory-discipline.zip - 2
Add it to your agent
Claude Code (all projects)
mkdir -p ~/.claude/skills && unzip agent-memory-discipline.zip -d ~/.claude/skills/Claude Code / Agent SDK (this project only)
mkdir -p .claude/skills && unzip agent-memory-discipline.zip -d .claude/skills/Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.
- 3
Ask for the task
No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.
Get better results
- Connect a memory backend first; the skill only tells the agent when to use it.
- Keep saved memories to one fact each so outdated facts can be closed without losing others.
- Do not store secrets or personal data; the skill says not to, but review what gets saved.
What's inside
- references/entry-format.md
- SKILL.md
Redistributed unmodified from sickn33/agentic-awesome-skills under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.