a-pavithraa/springboot-skills-marketplace / plugins/springboot-architecture/skills/spring-boot-team-harness
spring-boot-team-harness
VerifiedUse when a team wants every developer's coding agent (Claude Code or Codex) to follow the same rules and coding conventions on a Maven or Gradle Spring Boot 4 repository: setting up AGENTS.md, coding standards, guardrails, shared agent settings, or build and CI checks that agents must pass before code merges. It adds ./mvnw verify or ./gradlew build gates that fail on null-safety bugs (NullAway with JSpecify), module-boundary breaches (Spring Modulith), unknown or deprecated Boot property keys, Spring mistakes that compile (ArchUnit) and unformatted code, Claude Code and Codex hooks that run those gates before a turn ends, and a CI workflow.
assets/project/.claude/hooks/guard.sh
View on GitHub#!/usr/bin/env bash
# PreToolUse hook for Claude Code and Codex. Refuses actions the team never wants an agent to take: it prints a deny
# decision as JSON, which both tools honour and show to the agent. Codex 0.166 let an apply_patch edit through on exit 2.
#
# Each refusal appends one JSON line to .git/harness/events.jsonl (time, session, rule; no command or path), which the
# harness's report.py summarises.
set -u
input=$(cat)
# Codex sets no project folder variable and runs hooks in the session's folder, which can be a subfolder.
root=${CLAUDE_PROJECT_DIR:-$(git rev-parse --show-toplevel 2> /dev/null || pwd)}
rule_tests='CodeRulesTests|ConfigurationKeysTests|BuildSetupTests|UnusedCodeTests|ModuleBoundariesTests|ModularityTests|ArchUnitTests|ArchitectureRulesTest|LogLeakCheck'
# block <reason> <rule>
block() {
local dir session
if dir=$(cd "$root" 2> /dev/null && git rev-parse --absolute-git-dir 2> /dev/null); then
session=$(printf '%s' "$input" | grep -o '"session_id" *: *"[^"]*"' | head -1 | sed 's/.*"\([^"]*\)"$/\1/' \
| tr -cd 'A-Za-z0-9_-')
mkdir -p "$dir/harness" 2> /dev/null && printf '{"ts":"%s","hook":"guard","session":"%s","result":"refused","rule":"%s"}\n' \
"$(date -u +%Y-%m-%dT%H:%M:%SZ)" "${session:-unknown}" "$2" >> "$dir/harness/events.jsonl" 2> /dev/null
fi
printf '{"hookSpecificOutput":{"hookEventName":"PreToolUse","permissionDecision":"deny","permissionDecisionReason":"%s"}}\n' \
"$(printf '%s' "Blocked by the team harness: $1" | sed 's/\\/\\\\/g; s/"/\\"/g')"
exit 0
}
if printf '%s' "$input" | grep -q '"tool_name" *: *"Bash"'; then
command=$(printf '%s' "$input" | grep -oE '"command" *: *"(\\.|[^"\\])*"' | head -1 \
| sed -E 's/^"command" *: *"//; s/"$//; s/\\"/"/g; s/\\\\/\\/g')
printf '%s' "$command" | grep -Eq 'git[[:space:]].*push([[:space:]].*)?[[:space:]](--force|-f)([[:space:]]|$)' \
&& block "force-pushing rewrites shared history. Push normally, or ask the user to force-push themselves." force-push
printf '%s' "$command" | grep -Eq 'git[[:space:]].*reset([[:space:]].*)?[[:space:]]--hard' \
&& block "git reset --hard discards uncommitted work. Use git stash, or ask the user." reset-hard
printf '%s' "$command" | grep -Eq 'git[[:space:]].*clean([[:space:]].*)?[[:space:]]-[a-zA-Z]*f' \
&& block "git clean -f deletes untracked files for good. Ask the user." clean-force
printf '%s' "$command" | grep -Eq 'allowStoreCreation=true|freeze\.refreeze=true|modulith\.store\.record=true|harness\.record=true' \
&& block "that setting records today's ArchUnit or Spring Modulith violations as accepted. Fix the violation instead; to record a baseline on purpose, ask the user to run the harness's baseline.py." archunit-record
# Skipping is fine for packaging a jar to run; a gate run that skips checks proves nothing.
printf '%s' "$command" | grep -Eq '(^|[[:space:]])(verify|install|build|check)([[:space:]]|$)' \
&& printf '%s' "$command" | grep -Eq -- '-D(skipTests|maven\.test\.skip|jacoco\.skip|enforcer\.skip)(=true)?([^[:alnum:]._=-]|$)|(-x|--exclude-task)[[:space:]]+:?test([[:space:]]|$)' \
&& block "that runs the gates with tests or checks skipped, so a pass proves nothing. Run it without the skip flag, and fix what fails." skip-checks
# Writing commands that name a protected file. Not exhaustive: the Harness integrity CI check is the backstop.
printf '%s' "$command" | grep -Eq "archunit_store|modulith_store|archunit\.properties|package-info\.java|($rule_tests)\.java" \
&& printf '%s' "$command" | grep -Eq '(^|[^0-9&])>|(^|[[:space:];&|])(sed|perl)([[:space:]]+-[^[:space:]]+)*[[:space:]]+(-[a-zA-Z]*i|--in-place)|(^|[[:space:];&|])(tee|truncate|rm|mv|cp|python3?|node|ruby)([[:space:]]|$)' \
&& block "that command writes to a file holding the team's build rules or a record of tolerated violations. Fix the code the rule reports. To change a rule, ask the user." bash-protected-file
exit 0
fi
# check_path <path>: refuses an edit to a protected file, and returns when the edit is fine.
check_path() {
local path=$1 relative ref
case "$path" in
/* | [A-Za-z]:/*) ;;
*) path="$PWD/$path" ;;
esac
case "$path" in
*/archunit_store/*)
block "src/test/resources/archunit_store records the rule violations that existed before the harness, so they're frozen. Fix the new violation instead. To accept one on purpose, ask the user." archunit-store ;;
*/modulith_store)
block "src/test/resources/modulith_store records the module-boundary violations that existed before the harness, so they're tolerated. Fix the new violation instead. To accept one on purpose, ask the user." modulith-store ;;
*/archunit.properties)
block "archunit.properties sets how the freeze record works; with store creation allowed, new violations would be recorded as accepted. Fix the violation instead, or ask the user." archunit-properties ;;
*/CodeRulesTests.java | */ConfigurationKeysTests.java | */BuildSetupTests.java | */UnusedCodeTests.java | */ModuleBoundariesTests.java \
| */ModularityTests.java | */ArchUnitTests.java | */ArchitectureRulesTest.java | */LogLeakCheck.java)
[ -e "$path" ] && block "${path##*/} holds the team's build rules, and a change to it changes what every developer's build accepts. Fix the code the rule reports. To change a rule, ask the user." rule-test
return 0 ;;
*/package-info.java)
# A module's allowedDependencies and type are its boundary rules: widening them would get any breach past the
# boundary test. Telling a widening edit from a harmless one takes parsing Java, so every edit to a module's
# package-info.java is refused; creating one for a new module is fine.
[ -e "$path" ] && grep -q 'ApplicationModule' "$path" \
&& block "${path%/*}/package-info.java declares the module's boundary rules (allowedDependencies, OPEN). Call the other module through an allowed dependency, or ask the user to change the rule." module-rules
return 0 ;;
*/db/migration/*.sql) ;;
*) return 0 ;;
esac
relative=$(git -C "$root" ls-files --full-name -- "$path" 2> /dev/null)
[ -n "$relative" ] || return 0
for ref in origin/main origin/master main master; do
if git -C "$root" rev-parse --verify -q "$ref" > /dev/null; then
git -C "$root" cat-file -e "$ref:$relative" 2> /dev/null \
&& block "$relative is on $ref, so it may already have run on a shared database, and Flyway fails when its checksum changes. Write a new migration instead (see the db-migration skill)." applied-migration
return 0
fi
done
}
# Codex edits files with apply_patch: the patch text names each file it updates, deletes or moves to. Adding a file
# passes, as Write does for a file that doesn't exist yet.
if printf '%s' "$input" | grep -q '"tool_name" *: *"apply_patch"'; then
# The patch is a JSON string: an escaped backslash in a Windows path becomes /, then each \n ends a line.
printf '%s' "$input" | awk '{ gsub(/\\\\/, "/"); gsub(/\\n/, "\n"); print }' \
| sed -n -E 's/^\*\*\* (Update File|Delete File|Move to): (.*[^[:space:]])[[:space:]]*$/\2/p' \
| while IFS= read -r file; do
check_path "$file"
done
exit $?
fi
path=$(printf '%s' "$input" | grep -o '"file_path" *: *"[^"]*"' | head -1 | sed 's/.*: *"//; s/"$//; s#\\\\#/#g')
[ -n "$path" ] && check_path "$path"
exit 0
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |