How do I install the Coach skill?

Run npx skillmds@latest add netresearch/coach in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Coach safe to use?

This skill has not completed SkillMD's automated safety review yet. Independent scanners report: SkillSpector: PASS, Skill Scanner: PASS. Capability flags: executes scripts. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Coach?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Coach free to use?

Yes. Installing skills from SkillMD is free. This skill is licensed under (MIT AND CC-BY-SA-4.

netresearch (@netresearch) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Coach by netresearch — Agent Skill

Name: Coach
Rating: 4.5 (4 reviews)
Author: netresearch

File contents

Coach - Self-Improving Learning System

Coach enables Claude to learn from friction and improve over time. It detects learning opportunities (user corrections, repeated instructions, tool failures, tone escalation), extracts actionable improvement candidates, and proposes changes requiring explicit user approval.

Core Principle: No silent writes. All improvements require user approval via /coach approve.

Activation Triggers

Activate when: user corrections ("no", "stop", "don't"), repeated instructions, tool/command failures, tone escalation (ALL CAPS, "!!!"), skill supplements ("also remember..."), deprecated-tool warnings, explicit /coach commands, or session end.

Signal Categories (Priority Order)

COMMAND_FAILURE (Highest) - Non-zero exit, stderr patterns
USER_CORRECTION (High) - Explicit correction language
SKILL_SUPPLEMENT (High) - Additional guidance for a skill
VERSION_ISSUE (Medium-High) - Deprecated/outdated warnings
REPETITION (Medium) - Semantically similar instruction repeated
TONE_ESCALATION (Low) - Frustration indicators

Candidate Types

Type	Description	Example
`rule`	Stable constraint	"Never edit generated files"
`checklist`	Workflow step	"Run tests after code change"
`snippet`	Repeatable command	"Preflight check script"
`skill`	Skill update suggestion	"Add X guidance to Y skill"
`antipattern`	Things to never do	"Never assume tool exists"

Workflow Summary

Detection — hooks capture events → ~/.claude-coach/events.sqlite
Generation — aggregate signals into proposals (fingerprints dedupe)
Scope — project vs global per path/language
Review — /coach review approves/rejects/edits
Apply — approved rules → ~/.claude/CLAUDE.md (global) or repo AGENTS.md (project)
Retro — /coach retro maps manual work to skill gaps, opens PRs at source repos

File Locations

~/.claude-coach/
├── events.sqlite      # Raw friction events
├── candidates.json    # Pending proposals
└── ledger.sqlite      # Cross-repo fingerprints

~/.claude/CLAUDE.md     # Global rules destination
<repo>/AGENTS.md        # Project rules destination

~/.claude/ or <repo>/.claude/
├── checklists/        # Workflow checklists
└── snippets/          # Reusable commands

Scope

Coach: live in-session signal — events.sqlite capture, /coach commands. retro: session-end / cross-session analysis, skill-PR routing. /coach retro overlaps retro's job; merging them is an open decision.

Scripts

scripts/ (this dir) — self-contained for skill-dir installs. Plugin path: ${CLAUDE_PLUGIN_ROOT}/skills/coach/scripts/:

Script	Purpose
`init_coach.py`	Initialize coach system
`detect_signals.py`	Pattern detection for friction
`aggregate.py`	Turn signals into candidates
`skill_analyzer.py`	Analyze skills and scan for outdated tools
`apply.py`	Apply approved proposals

Proactive Commands

/coach scan — Check for outdated CLI tools and dependencies
/coach retro — Session retrospective: analyze manual work, create PRs to improve skills at source

For detailed architecture, schemas, and patterns, see references/ directory.

---name: coachdescription: Use when Claude makes repeated mistakes, users correct behavior, tool failures indicate missing knowledge, or managing learning via /coach commands.license: (MIT AND CC-BY-SA-4.0). See LICENSE-MIT and LICENSE-CC-BY-SA-4.0---# Coach - Self-Improving Learning SystemCoach enables Claude to learn from friction and improve over time. It detects learning opportunities (user corrections, repeated instructions, tool failures, tone escalation), extracts actionable improvement candidates, and proposes changes requiring explicit user approval.**Core Principle**: No silent writes. All improvements require user approval via `/coach approve`.## Activation TriggersActivate when: user corrections ("no", "stop", "don't"), repeated instructions, tool/command failures, tone escalation (ALL CAPS, "!!!"), skill supplements ("also remember..."), deprecated-tool warnings, explicit `/coach` commands, or session end.## Signal Categories (Priority Order)1. **COMMAND_FAILURE** (Highest) - Non-zero exit, stderr patterns2. **USER_CORRECTION** (High) - Explicit correction language3. **SKILL_SUPPLEMENT** (High) - Additional guidance for a skill4. **VERSION_ISSUE** (Medium-High) - Deprecated/outdated warnings5. **REPETITION** (Medium) - Semantically similar instruction repeated6. **TONE_ESCALATION** (Low) - Frustration indicators## Candidate Types| Type | Description | Example ||------|-------------|---------|| `rule` | Stable constraint | "Never edit generated files" || `checklist` | Workflow step | "Run tests after code change" || `snippet` | Repeatable command | "Preflight check script" || `skill` | Skill update suggestion | "Add X guidance to Y skill" || `antipattern` | Things to never do | "Never assume tool exists" |## Workflow Summary1. **Detection** — hooks capture events → `~/.claude-coach/events.sqlite`2. **Generation** — aggregate signals into proposals (fingerprints dedupe)3. **Scope** — project vs global per path/language4. **Review** — `/coach review` approves/rejects/edits5. **Apply** — approved rules → `~/.claude/CLAUDE.md` (global) or repo `AGENTS.md` (project)6. **Retro** — `/coach retro` maps manual work to skill gaps, opens PRs at source repos## File Locations```~/.claude-coach/├── events.sqlite # Raw friction events├── candidates.json # Pending proposals└── ledger.sqlite # Cross-repo fingerprints~/.claude/CLAUDE.md # Global rules destination<repo>/AGENTS.md # Project rules destination~/.claude/ or <repo>/.claude/├── checklists/ # Workflow checklists└── snippets/ # Reusable commands```## ScopeCoach: live in-session signal — `events.sqlite` capture, `/coach` commands.`retro`: session-end / cross-session analysis, skill-PR routing. `/coachretro` overlaps `retro`'s job; merging them is an open decision.## Scripts`scripts/` (this dir) — self-contained for skill-dir installs. Plugin path:`${CLAUDE_PLUGIN_ROOT}/skills/coach/scripts/`:| Script | Purpose ||--------|---------|| `init_coach.py` | Initialize coach system || `detect_signals.py` | Pattern detection for friction || `aggregate.py` | Turn signals into candidates || `skill_analyzer.py` | Analyze skills and scan for outdated tools || `apply.py` | Apply approved proposals |## Proactive Commands- `/coach scan` — Check for outdated CLI tools and dependencies- `/coach retro` — Session retrospective: analyze manual work, create PRs to improve skills at sourceFor detailed architecture, schemas, and patterns, see `references/` directory.

Frequently asked questions

Coach