Reflect

Battle-tested Agent Skills for Claude Code — ship → babysit pipeline and friends. Distributed via positronick.com

Install

npx -y skills add nsollazzo/skills --skill reflect

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

0 stars0 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 author says it does

Copied from the file, not written here

Capture session knowledge (gotchas, patterns, integration quirks) into the project's knowledge base. Use when: session involved tricky bugs, new integrations, non-obvious patterns, or hard-won lessons. Also use when asked to "reflect", "capture lessons", "what did we learn", "session recap", or "save knowledge". Works after any coding session, not just commits.

The file declares its own license as MIT. 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

3.6 KB, 785 tokens by cl100k_base, as published. Nobody here has run it

Reflect — Session Knowledge Capture

Scan the current session for valuable lessons and route them to the right knowledge target.

Workflow

Step 1: Gather Session Context

Read-only scan of what happened this session:

git diff main --stat
git log main..HEAD --oneline
git diff main

Also review conversation context: problems solved, errors hit, workarounds applied, patterns discovered.

Step 2: Scan Existing Knowledge Base

Quick skim (not full read) to avoid duplicates:

~/.claude/projects/*/memory/MEMORY.md — current gotchas
Project CLAUDE.md — current rules
~/.claude/skills/ directory listing — skill coverage
docs/ directory listing — existing documentation

Step 3: Identify Knowledge Gaps

Classify 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:

Skip routine changes (simple renames, formatting, obvious fixes)
Skip duplicates of existing knowledge
Max 7 suggestions
Priority: gotchas > rules > skill updates > docs > new skills

Step 4: Present Summary

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.

Step 5: User Approval

Use AskUserQuestion with multiSelect. Options:

Each numbered suggestion as a selectable item
"Apply all"
"Skip all"

NEVER auto-apply without explicit user approval.

Step 6: Apply Approved Updates

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.

Step 7: Commit Knowledge Changes (Optional)

Ask user whether to commit with: docs: capture session lessons

If user declines, leave changes staged but uncommitted.

Rules

NEVER auto-apply without user approval (Step 5)
NEVER create new skills directly — route to /skill-creator
NEVER modify source code — only knowledge files (MEMORY.md, CLAUDE.md, skills, docs)
Max 7 suggestions per reflection
Keep it FAST — this should take under 60 seconds
Gotchas → MEMORY.md, permanent patterns → CLAUDE.md
One sentence of "why" context per suggestion — no essays