skills.rest
lucasheriques
lucasheriquesCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

reflect

Spawns three parallel review subagents over a session transcript to route learnings into skill edits.

Updated Sep 23, 2026
Search Tags:#code review#claude code#subagents#continuous learning#transcript analysis#reflection#skill improvement
One-click install
npx skills add https://github.com/lucasheriques/pstack --skill reflect-lucasheriques
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: reflect
Source: https://github.com/lucasheriques/pstack/tree/main/skills/reflect
Command: npx skills add https://github.com/lucasheriques/pstack --skill reflect-lucasheriques

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Long engineering sessions produce durable lessons that are lost when the conversation ends. This Skill mines the active transcript for those learnings and converts them into concrete, reviewed edits to existing skills so future agents benefit. ## Core Features & Use Cases - Parallel multi-lens review: Spawns three read-only subagents (judgment, tooling, divergent) over the session transcript, each with a dedicated prompt template. - Synthesis and triage: A synthesizer agent merges findings into an Accepted / Rejected / Backlog list with durability, specificity, and convergence checks. - Approval-gated application: Accepted edits are applied only after explicit user approval, routed to direct edits, the skill-creator skill, or description tuning. - Use Case: After a long debugging session where the agent learned a lockfile quirk and missed a skill trigger, run reflect to surface those lessons and patch the relevant skills. ## Quick Start Ask the agent to reflect on this session and turn what it learned into skill improvements.

Frequently Asked Questions about reflect

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I turn lessons from a Claude Code session into skill improvements?▼

Invoke the reflect skill after a session. It locates the active transcript, spawns three parallel review subagents with different lenses, synthesizes their findings, and presents an approval-gated list of edits to existing skills.

What does the reflect skill review in a session transcript?▼

It reviews mistakes and corrections, tool and library quirks, missed skill triggers, second-order effects, and moments where the user manually supplied context the agent could have fetched itself via MCP tools.

Does reflect modify skills automatically without approval?▼

No. The synthesizer produces an Accepted, Rejected, and Backlog list, and the parent agent presents it to the user and waits for explicit approval before applying any edit. Only backlog items are filed automatically.

Can reflect create new skills if no existing skill fits a learning?▼

Yes, but only when no existing skill is a real home and the pattern recurs. New skills are routed to the skill-creator skill rather than being drafted ad hoc, keeping the structure consistent.

When should I not run reflect on a session?▼

Skip it when the conversation is trivial, off-topic, or already covered by an existing skill the agent followed correctly. One-off incidents are not durable learnings and are filtered out by the reviewers.

Related Skills

Explore More Software Engineering Skills→
amdamd
1.6k

analyzing-claude-sessions

Analyze local Claude Code session transcripts to measure usage patterns, failures, and token costs.

Official
Advanced
ShawhinTShawhinT
40

skill-updater

Update and repackage existing Claude skills from a writable working directory.

Community
Intermediate
ronniegeraghtyronniegeraghty
3

skill-authoring

Validate SKILL.md frontmatter and body against the agentskills.io spec.

Community
Intermediate
koseki2580koseki2580

writing-skills

Create and edit Skill documentation using a RED-GREEN-REFACTOR workflow.

Community
Advanced