Memory
Per-role persistent memory for Octobots roles. Curated typed entries (facts, preferences, decisions) and append-only daily logs (episodic, "what did I do today"). Use when the user says "remember this", "what did you learn yesterday", "log this", or whenever you discover something worth keeping across sessions.From its SKILL.md
npx -y skills add arozumenko/wikis --skill memoryAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 6 stars6 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its file declares
Copied from the file, not written here
The file declares its own license as Apache-2.0. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.
SKILL.md
3.2 KB, as published. Nobody here has run it
Memory
Persistent per-role memory. Two complementary stores under .octobots/memory/<role>/:
- Curated — typed entries (
<name>.mdwith frontmatter) plus an indexMEMORY.md. For facts that should survive across many sessions: user preferences, project decisions, feedback, references. - Daily log —
daily/YYYY-MM-DD.md, append-only timestamped lines. For episodic recall: what you did, what the user said, transient context. Bounded by date so it doesn't grow forever.
A snapshot.md is regenerated at every role launch combining the index and the last few days of daily log. Roles @import it from their AGENT.md so memory is auto-loaded into the system prompt at session start — no manual reads needed.
Quick Start
Resolve your role from env OCTOBOTS_ID.
# Append a timestamped line to today's daily log
python octobots/skills/memory/scripts/memory.py log "Discussed Q2 roadmap with user; agreed to defer mobile work"
# Write a curated typed memory entry (use for things worth months/years)
python octobots/skills/memory/scripts/memory.py write user_timezone \
--type user --description "User's timezone and quiet hours" \
--content "User is in CET (UTC+1). Quiet hours: 22:00–08:00."
# Read the last 3 days of daily logs + curated index
python octobots/skills/memory/scripts/memory.py read --days 3
# Regenerate snapshot.md (supervisor calls this automatically at launch)
python octobots/skills/memory/scripts/memory.py snapshot
All commands accept --role <name> to override OCTOBOTS_ID.
When to log vs. write
log— anything happening now that you'd want to remember tomorrow but probably not next month. User's mood, what task you're mid-flight on, a one-off observation. Cheap.write— durable facts, preferences, decisions, lessons. Expensive — costs an index entry. Should be re-readable in 6 months and still useful.
If unsure: log it. The next snapshot will surface it; you can promote to a curated entry later.
Memory types (curated only)
| Type | What goes here |
|---|---|
user | Who the user is, role, expertise, preferences |
feedback | Corrections and validated approaches the user gave you. Include the why |
project | Goals, deadlines, constraints, in-flight initiatives. Decay fast — re-verify before acting |
reference | Pointers to external systems (Linear projects, Slack channels, dashboards) |
Auto-load
Roles import their snapshot via @.octobots/memory/<role>/snapshot.md at the top of their AGENT.md. Claude Code resolves the import at session start so curated entries and recent daily logs are part of the system prompt automatically. The supervisor regenerates snapshot.md before every role spawn.