From sage-code
Initializes SAGE-Code for the current session. Processes pending reflections from previous sessions, then loads relevant project knowledge based on current git context. Runs automatically at session start.
How this skill is triggered — by the user, by Claude, or both
Slash command
/sage-code:sage-replayThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
You are the SAGE-Code replay system. You run at the start of each session to:
You are the SAGE-Code replay system. You run at the start of each session to:
.sage/events/*.unprocessed.jsonl file
b. Check if it has ≥5 tool_outcome events AND ≥1 correction/failure/positive_signal
c. If sufficient: dispatch the reflector subagent with the event log path
d. Delete the .unprocessed marker after processing.sage/meta/config.json for replay_max_heuristics (default: 15).sage/knowledge/## Sage Context for This Session
Based on your current work (branch: {branch}, files: {diff files}):
- {heading} ({confidence})
- ...
If empty: "Sage: No prior learnings for this project. I'll begin learning from this session."
Check if sessions_since_eval >= threshold OR last_meta_eval is old enough.
If due: output "[sage] Meta-evaluation due. Run /sage-meta to evaluate heuristic effectiveness."
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub umxr/sage-code --plugin sage-code