reflect — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited reflect (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Scan the current session for valuable lessons and route them to the right knowledge target.
Read-only scan of what happened this session:
git diff main --stat
git log main..HEAD --oneline
git diff mainAlso review conversation context: problems solved, errors hit, workarounds applied, patterns discovered.
Quick skim (not full read) to avoid duplicates:
~/.claude/projects/*/memory/MEMORY.md — current gotchasCLAUDE.md — current rules~/.claude/skills/ directory listing — skill coveragedocs/ directory listing — existing documentationClassify each insight into exactly one category:
| Category | Target | When |
|---|---|---|
| Gotcha | MEMORY.md | Non-obvious trap or workaround |
| Rule | CLAUDE.md | Project-wide always/never pattern |
| Skill update | .claude/skills/<name>/ | Existing skill missing info |
| New skill | /skill-creator | New integration worth codifying |
| Docs | docs/ | Missing guide or troubleshooting |
Filters:
Concise table — scannable in 5 seconds:
## Session Reflection
| # | Category | Target | Suggestion |
|---|----------|--------|------------|
| 1 | Gotcha | MEMORY.md | Convex v.array() inside unions panics |
| 2 | Skill | /workos | Add headless login refresh pattern |
**Why 1:** We hit a runtime panic when... (1 sentence)
**Why 2:** The headless flow required... (1 sentence)If nothing worth capturing: "No new knowledge. Existing docs cover this session." — stop here.
Use AskUserQuestion with multiSelect. Options:
NEVER auto-apply without explicit user approval.
For each approved suggestion:
| Category | Action |
|---|---|
| Gotcha | Append to MEMORY.md under appropriate section heading |
| Rule | Add to CLAUDE.md in the relevant section |
| Skill update | Edit the skill's SKILL.md or references/ |
| New skill | Print "Run /skill-creator to build X" — NEVER auto-create skills |
| Docs | Create or update file in docs/ |
Keep edits minimal — add bullet points, don't reorganize existing content.
Ask user whether to commit with: docs: capture session lessons
If user declines, leave changes staged but uncommitted.
/skill-creator~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.