jordimarsal/harness-standard
A standardized multi-agent harness for Opencode and Claude Code. Installs into any project with a single command.
ARCHETYPE
Pragmatist
A balanced, no-drama setup: some rules, some tools, nothing extreme.
Copy this rig
# review before running: this installs third-party code
$ npx degit jordimarsal/harness-standard/templates/.claude ./rig-harness-standard # inspect, then merge into .claude/ MCP servers are added to Claude Code at local scope; env vars are shown as YOUR_… placeholders — we never store values. Files are fetched with degit into a separate folder so you can review before merging.
$ curl -fsSL --create-dirs -o .claude/agents/implementer.md https://raw.githubusercontent.com/jordimarsal/harness-standard/main/templates/.claude/agents/implementer.md $ curl -fsSL --create-dirs -o .claude/agents/leader.md https://raw.githubusercontent.com/jordimarsal/harness-standard/main/templates/.claude/agents/leader.md $ curl -fsSL --create-dirs -o .claude/agents/reviewer.md https://raw.githubusercontent.com/jordimarsal/harness-standard/main/templates/.claude/agents/reviewer.md $ curl -fsSL --create-dirs -o .claude/agents/spec-author.md https://raw.githubusercontent.com/jordimarsal/harness-standard/main/templates/.claude/agents/spec-author.md
This rig commits no guardrails. Here is the community baseline instead — the deny/ask rules most often found across all 7,204 rigs:
{
"permissions": {
"deny": [
"Read(./.env)",
"Read(**/.env)",
"Read(~/.ssh/**)",
"Bash(rm -rf *)",
"Read(**/*.pem)",
"Bash(rm -rf /)",
"Bash(git push --force:*)",
"Bash(sudo *)",
"Read(.env)",
"Bash(rm -rf /*)",
"Read(./.env.*)",
"Read(~/.aws/**)",
"Bash(git push --force*)",
"Bash(rm -rf:*)",
"Read(**/*.key)",
"Read(**/.env.*)",
"Bash(sudo:*)",
"Bash(git reset --hard*)",
"Bash(git reset --hard:*)",
"Read(.env.*)"
],
"ask": [
"Bash(git push:*)",
"Bash(git push *)",
"Bash(git commit:*)",
"Bash(rm *)",
"Bash(rm:*)",
"Bash(git rebase *)",
"Bash(wget *)",
"Bash(npm publish:*)",
"Bash(git commit *)",
"Bash(gh pr merge *)"
]
}
} Subagents (4)
| implementer | Worker. Implements exactly ONE feature from its approved spec. Writes code, writes tests, self-verifies. |
| leader | Orchestrator. Decomposes work and dispatches subagents. NEVER writes application code directly. |
| reviewer | Automated reviewer. Approves or rejects work against docs/, harness/specs/<name>/, and harness/CHECKPOINTS.md. |
| spec-author | Writes Kiro-style specs (requirements/design/tasks) for a pending feature. NEVER writes application code or tests. |
Hooks (2)
| event | matcher | runs |
|---|---|---|
| PostToolUse | Edit|Write | {{TEST_CMD}} 2>&1 | tail -3 |
| Stop | * | LOG=$(mktemp "${TMPDIR:-/tmp}/harness_init.XXXXXX") && harness/init.sh > "$LOG" 2>&1 && echo "[harness] init.sh OK ($LOG)" || (echo "[harness] init.sh FAILED - check $LOG before closing" && tail -20 "$LOG") |
Permissions
deny (0)
—
ask (0)
—
allow (3)
Bash(harness/init.sh)
Bash({{TEST_CMD}})
Bash({{BUILD_CMD}})
Similar rigs
0xT4sty/claude-code-dotfiles
—
Pragmatist 979 tok ·
Saba-Var/claude-ward
A read-only tripwire that watches Claude Code's local config and reports suspicious changes.
Pragmatist 1.8k tok ·
OMOCHInoHOSHI/dotfiles-claude
.claude
Pragmatist 293 tok ·
MyrskyDev/claude-agent-template
Multi-agent dev workflow template for Claude Code
Pragmatist 195 tok ·