Gptme Context

gptme:context

Load gptme workspace knowledge (lessons, skills, memory) into this Claude Code session for use in subsequent tasks.

When to Use

• User invokes /gptme:context
• User wants to leverage gptme workspace memory or lessons in this CC session
• Working in a project that has a gptme.toml and lessons/ directory

Procedure

From a gptme workspace (recommended)

If the current project has a lessons/ directory and gptme.toml:

LESSONS_DIR="$(git rev-parse --show-toplevel 2>/dev/null)/lessons"
if [ -d "$LESSONS_DIR" ]; then
  echo "=== gptme Lessons ==="
  find "$LESSONS_DIR" -name "*.md" -not -name "README.md" | \
    head -20 | xargs cat 2>/dev/null | head -3000
else
  echo "No lessons/ directory found in current workspace"
fi

From gptme global config

GLOBAL_LESSONS="${HOME}/.config/gptme/lessons"
if [ -d "$GLOBAL_LESSONS" ]; then
  echo "=== Global gptme Lessons ==="
  find "$GLOBAL_LESSONS" -name "*.md" | head -10 | xargs cat | head -2000
fi

List available skills

# List gptme skills if gptme-util is available
if command -v gptme-util &>/dev/null; then
  gptme-util skills list 2>/dev/null
fi

# Or list directly
SKILLS_DIR="$(git rev-parse --show-toplevel 2>/dev/null)/skills"
[ -d "$SKILLS_DIR" ] && find "$SKILLS_DIR" -name "SKILL.md" | head -20 | while read f; do
  echo "$(basename $(dirname $f)): $(head -5 $f | grep description | cut -d: -f2-)"
done

After loading context

Once injected, refer to the loaded lessons and skills naturally in subsequent prompts. Example: "Now apply the lessons you just loaded to review this code."

Notes

• Context is injected into this session only — it does not persist after the session ends
• For persistent behavior, install a gptme workspace with the relevant lessons/ and skills/ directories
• The Bob agent (https://github.com/ErikBjare/bob) is an example of a rich gptme workspace

Related

• gptme docs
• gptme agent template — fork to create your own workspace
• /gptme:run — delegate tasks to gptme
• /gptme:review — code review via gptme