DOCS

Agent Skills

Install the `pajama` CLI, then install public skills that teach agents how to use this memory system end-to-end.

Agent Skills

A skill is a folder that contains a SKILL.md (plus optional scripts/references) that teaches an agent how to do a workflow end-to-end.

In this repo, skills live under:

skills/<skill-name>/

To make a skill public, keep it in a public GitHub repo and make sure it contains no secrets.

Install The CLI First

Most of our skills assume you can call the CLI:

npm i -g @pajamadot/pajama
pajama login

Install A Skill (Codex / Claude Code)

The default skill directory is:

  • Windows: %USERPROFILE%\\.codex\\skills
  • macOS/Linux: ~/.codex/skills

PowerShell (Windows)

Installs the pajama-cli skill:

$CODEX = if ($env:CODEX_HOME) { $env:CODEX_HOME } else { Join-Path $HOME ".codex" }
$tmp = Join-Path $env:TEMP "gdm-skill"
if (Test-Path $tmp) { Remove-Item -Recurse -Force $tmp }
git clone --depth 1 https://github.com/pajamadot/game-dev-memory.git $tmp
New-Item -ItemType Directory -Force (Join-Path $CODEX "skills\\pajama-cli") | Out-Null
Copy-Item -Recurse -Force (Join-Path $tmp "skills\\pajama-cli\\*") (Join-Path $CODEX "skills\\pajama-cli")

Bash (macOS/Linux)

CODEX_HOME="${CODEX_HOME:-$HOME/.codex}"
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/pajamadot/game-dev-memory.git "$tmp/gdm"
mkdir -p "$CODEX_HOME/skills/pajama-cli"
cp -R "$tmp/gdm/skills/pajama-cli/"* "$CODEX_HOME/skills/pajama-cli/"

Bundled Skills

SkillWhat it's for
pajama-cliUse pajama to record memories and attach evidence assets (logs/zips/traces).
memory-evolverSelf-evolution loop that reads /api/evolve/signals and writes evolution events and seed knowledge.
unreal-agentsUnreal Engine + AI agent research workflows (and daily digest).
agent-memory-researchResearch and notes on memory organization patterns and schema design.
ux-e2eEnd-to-end UX testing guidance (Playwright).

A Good Default Workflow For Agents

  1. Create a memory for what happened and why.
  2. Upload evidence files as assets and link them to the memory.
  3. Run evolver cycles (optional) when sessions close or when the system is stable.