- Home
- Skills
- Agents & Orchestration
- Agent Memory Journal: Plain-Markdown Memory for Any Agent
More screenshots
Works with the AI Agents you already use
Agent Memory Journal: Plain-Markdown Memory for Any Agent
Agents forget everything between sessions, and the built-in memory some tools offer is locked to one tool, hidden in a home folder, and hard to review.
$7
Agent Memory Journal: Plain-Markdown Memory for Any Agent
Example session with this skill installed
We moved staging off Heroku to Fly.io last Thursday. Staging
deploys are now fly deploy --config fly.staging.toml, so remember that.
Also forget the Friday deploy freeze, the team dropped it. One gotcha I hit:
fly deploy quietly uses fly.toml (production) unless you pass --config, so
note that for next time. Then log this session.
- Read your context and instructions
- Compiled the agent memory journal
I've updated the memory files in agent-memory/, and the checker reports
no problems.
- Updated
memory/deploy-staging.md: staging is on Fly.io, and the
deploy command isfly deploy --config fly.staging.toml. I recorded the
move date as 2026-11-05 (last Thursday). It also says to always pass
--config, because without it the deploy goes to production. I changed
its line inMEMORY.mdto match. - Deleted
memory/friday-freeze.mdand removed its line from
MEMORY.md. No other memories linked to it. - Added the
fly deploy/fly.tomlgotcha tolearnings.md. - Logged this session as the first entry in a new file,
journal/2026-11.md.
One older learning in learnings.md is about Heroku review apps. Staging
isn't on Heroku anymore, but production may still be, so I kept it. Tell me
if it's out of date and I'll remove it.
What changed on disk:
MEMORY.md
- [Deploy staging](...) — staging on Heroku, git push heroku-staging
- [Friday deploy freeze](...) — no prod deploys after Fri 15:00
+ [Deploy staging](...) — Fly.io: fly deploy --config fly.staging.toml
journal/2026-11.md (new)
Connects securely to your tools. The creator never sees your data.
About this skill
Agents forget everything between sessions, and the built-in memory some tools offer is locked to one tool, hidden in a home folder, and hard to review. Agent Memory Journal gives your agent a memory you can open in any editor, commit to git and share between agents and models.
What it sets up. One folder, agent-memory/:
MEMORY.md: the index, one line per memory, read at the start of a session. Short by design, so it costs little context.memory/<slug>.md: one fact per file, with a one-line description and a type:user(who you are),feedback(how you want the agent to work),project(decisions, goals, deadlines) orreference(where things live). Rules and decisions carry Why: and How to apply: lines, so the agent knows when a rule fits.journal/<YYYY-MM>.md: one short dated entry per session you ask it to log.learnings.md: reusable gotchas and commands, deduplicated and dated.
The rules that make it work. The skill tells the agent exactly when to save, update and delete. It checks for an existing memory before writing, so a changed fact updates one file instead of piling up contradictions. It turns "last Thursday" into a real date. It removes the index line along with the file when you say "forget". It never stores secret values, only where a secret lives. It skips what the code or git history already shows. Before acting on a memory that names a file or command, it checks that the thing still exists.
A checker included. scripts/check_memory.py (Python 3 standard
library, read-only) reports index lines pointing at missing files, memories
missing from the index, bad frontmatter, unwritten [[links]], repeated
learnings and strings that look like secrets.
No hooks. The skill triggers when you say "remember", "forget", "what do you remember" or "log this session". To load the index every session, add one line to your AGENTS.md or CLAUDE.md; the skill offers the line the first time.
Requires: an agent that loads SKILL.md skills and can read and write files, such as Claude Code or another Agent Skills compatible agent. Python 3 is optional, only for the checker. No API keys, no network.
Known limitations
- The agent reads the index when the skill triggers or your instructions tell it to. Without the one-line pointer in AGENTS.md or CLAUDE.md, a session that never mentions memory won't open it.
- It is a file convention, not a search engine. It suits hundreds of memories, not hundreds of thousands.
- If your agent also has its own built-in memory, both exist side by side.
The skill writes to
agent-memory/whenever the folder exists or you ask for it. - The secret check is pattern-based. It catches common key formats, not every secret.
FAQ
Does it work with models other than Claude? Yes. It is plain markdown and plain rules; any agent that reads skills or follows an instructions file can use it.
Can my team share one memory? Yes. Commit agent-memory/ to the repo.
Personal preferences fit better in a folder of your own; name it and the
skill uses it.
What if a memory is wrong? Say "forget X" or "that's out of date". The agent deletes or rewrites the file and its index line and tells you what changed.
Is anything sent anywhere? No. Files stay on your disk.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Agents & Orchestration.
The agent reads the index when the skill triggers or your instructions tell it to. Without the one-line pointer in AGENTS.md or CLAUDE.md, a session that never mentions memory won't open it. It is a file convention, not a search engine. It suits hundreds of memories, not hundreds of thousands. If your agent also has its own built-in memory, both exist side by side. The skill writes to agent-memory/ whenever the folder exists or you ask for it. The secret check is pattern-based. It catches common key formats, not every secret.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 9 days ago
- Software
Git