~ / rigs / aeiadev / harness

aeiadev/harness

Session context, command guard, nine roles and spend display for Claude Code and Codex

↗ GitHub ★ 0 MIT updated 1mo ago personal setup Claude Code Claude plugin
field: unknown workflow: Testing & TDDPlanning & specs
stack: Python
share on X
ARCHETYPE
Orchestrator
A bench of specialised subagents. The main agent mostly delegates.
CONTEXT TAX · EVERY TURN
~650 tokens
Featherweight · median rig: 2.3k · breakdown
GUARDRAILS
0/5
No committed guardrails · details
Quality 0/100 Guardrails v2

Copy this rig

# review before running: this installs third-party code
$ npx degit aeiadev/harness/agents ./rig-harness/agents
$ npx degit aeiadev/harness/skills ./rig-harness/skills
$ npx degit aeiadev/harness/hooks ./rig-harness/hooks

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.

$ npx degit <redacted> .claude/skills/checkpoint
$ npx degit aeiadev/harness/plugins/harness/skills/project-handoff .claude/skills/project-handoff
$ npx degit aeiadev/harness/plugins/harness/skills/skill-index .claude/skills/skill-index
$ npx degit aeiadev/harness/plugins/harness/skills/tdd .claude/skills/tdd
$ npx degit aeiadev/harness/plugins/harness/skills/verify-patch-landed .claude/skills/verify-patch-landed
$ curl -fsSL --create-dirs -o .claude/agents/builder-in-place.md https://raw.githubusercontent.com/aeiadev/harness/main/agents/builder-in-place.md
$ curl -fsSL --create-dirs -o .claude/agents/builder.md https://raw.githubusercontent.com/aeiadev/harness/main/agents/builder.md
$ curl -fsSL --create-dirs -o .claude/agents/docs-writer.md https://raw.githubusercontent.com/aeiadev/harness/main/agents/docs-writer.md
$ curl -fsSL --create-dirs -o .claude/agents/judge.md https://raw.githubusercontent.com/aeiadev/harness/main/agents/judge.md
$ curl -fsSL --create-dirs -o .claude/agents/planner.md https://raw.githubusercontent.com/aeiadev/harness/main/agents/planner.md
$ curl -fsSL --create-dirs -o .claude/agents/researcher.md https://raw.githubusercontent.com/aeiadev/harness/main/agents/researcher.md
$ curl -fsSL --create-dirs -o .claude/agents/sweeper.md https://raw.githubusercontent.com/aeiadev/harness/main/agents/sweeper.md
$ curl -fsSL --create-dirs -o .claude/agents/test-writer.md https://raw.githubusercontent.com/aeiadev/harness/main/agents/test-writer.md
$ curl -fsSL --create-dirs -o .claude/agents/worker.md https://raw.githubusercontent.com/aeiadev/harness/main/agents/worker.md

This rig commits no guardrails. Here is the community baseline instead — the deny/ask rules most often found across all 16.2k rigs:

{
  "permissions": {
    "deny": [
      "Read(./.env)",
      "Read(./.env.*)",
      "Read(**/.env)",
      "Read(~/.ssh/**)",
      "Bash(git push --force:*)",
      "Read(**/*.pem)",
      "Read(.env)",
      "Read(~/.aws/**)",
      "Read(**/*.key)",
      "Read(**/.env.*)",
      "Bash(git push --force*)",
      "Bash(rm -rf:*)",
      "Bash(rm -rf *)",
      "Bash(sudo *)",
      "Bash(sudo:*)",
      "Bash(rm -rf /)",
      "Bash(git push -f:*)",
      "Read(.env.*)",
      "Bash(git reset --hard*)",
      "Bash(rm -rf /*)"
    ],
    "ask": [
      "Bash(git push:*)",
      "Bash(git push *)",
      "Bash(gh pr merge *)",
      "Bash(git commit:*)",
      "Bash(rm *)",
      "Bash(gh pr merge:*)",
      "Bash(gh pr create *)",
      "Bash(git reset --hard:*)",
      "Bash(npm publish:*)",
      "Bash(git rebase *)"
    ]
  }
}

Skills (5)

Subagents (9)

builder-in-place
model: sonnet
Use to implement one bounded change from a TASK, FILES, BAR, RETURN brief in the current directory, when the caller already selected the checkout. Runs the check and reports.
builder
model: sonnet
Use to implement one bounded change from a TASK, FILES, BAR, RETURN brief in an isolated git worktree. Runs the check and reports.
docs-writer
model: sonnet
Use when a README or other docs need writing or updating from the actual code. Verifies examples within the named doc paths.
judge
model: opus
Use after an implementation to verify it in a fresh context against its brief. Reruns the check and returns a PASS or SEND_BACK verdict.
planner
model: sonnet
Use before implementation for a design critique or an executable plan. Reads the code and returns scoped steps and checks. Read-only.
researcher
model: sonnet
Use when a decision needs evidence or a comparison of options. Researches sources and recommends a choice. Read-only.
sweeper
model: haiku
Use to find files, list matches, count items, or summarize what exists within a named scope. Read-only; returns locations and counts.
test-writer
model: sonnet
Use when a named behavior needs a failing test first. Writes focused tests in the allowed paths and explains the observed failure.
worker
model: sonnet
Use for one multi-step workflow when a narrower role does not fit. Completes the outcome, verifies it, and reports remaining work.

Similar rigs

copied ✓