Agent builder

Skill aAAaqwq/AGI-Super-Team/skills/agent-builder

Build agent from spec: code, skill, config, launchdFrom its SKILL.md

Install

npx -y skills add aAAaqwq/AGI-Super-Team --skill agent-builder

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

SKILL.md

3.2 KB, 848 tokens by cl100k_base, as published. Nobody here has run it

Agent Builder

Takes a spec from Process Analyst and implements the agent: code, skill, config, launchd.

When to use

After Process Analyst has created a spec
"build an agent for process X"
"implement spec Y"

Input

Spec file from $AGENTS_PATH/specs/[name].spec.md

How to execute

Step 1: Read the spec

Read the spec file completely
Read the reference implementation: Email Pipeline ($GOOGLE_TOOLS_PATH/email_agent.py)
Understand the pipeline: trigger → steps → output

Step 2: Define architecture

Based on the spec, define:

agents/[name]/
├── [name]_agent.py        ← Main agent script
├── config.json            ← Configuration (paths, params)
├── README.md              ← Documentation
└── test_[name].py         ← Tests

Build rules:

One file = one step (if step is complex) or one file = entire pipeline (if simple)
Claude CLI for AI — use claude -p --model [model] instead of API key
CSV for data — read/write via pandas or csv module
Git auto-commit — if agent modifies CRM/PM data
Telegram notification — if human approval is needed
Dry-run mode — mandatory --dry-run flag
Logging — stdout for launchd, file for debug
Idempotency — re-run must not duplicate data

Step 3: Build

For each step from the spec:

Write the function/script
Handle errors according to the spec
Add logging
Add dry-run branch

Step 4: Create skill

Create skill file skills/agents/[name]-run.md with instructions on how to run the agent manually.

Step 5: Create launchd plist (if scheduled)

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "...">
<plist version="1.0">
<dict>
    <key>Label</key>
    <string>com.yourcompany.[name]-agent</string>
    <key>ProgramArguments</key>
    <array>
        <string>/usr/bin/python3</string>
        <string>$AGENTS_PATH/[name]/[name]_agent.py</string>
    </array>
    <key>StartInterval</key>
    <integer>[seconds]</integer>
    <key>StandardOutPath</key>
    <string>/tmp/[name]-agent.log</string>
    <key>StandardErrorPath</key>
    <string>/tmp/[name]-agent-error.log</string>
</dict>
</plist>

Step 6: Hand off to Agent Tester

Notify that the agent is ready for testing.

Output

Agent code in $AGENTS_PATH/[name]/
Skill file in $SKILLS_PATH/skills/agents/
Launchd plist (if scheduled)

Examples

Reference: Email Pipeline

google-tools/
├── email_monitor.py        ← Step 1: Gmail API check
├── email_agent.py          ← Step 2: AI classify (haiku)
├── email_action_agent.py   ← Step 3: CRM match + log
└── data/
    ├── email_summaries/    ← Output: summaries
    └── email_drafts/       ← Output: draft replies

Trigger: launchd every 3600s Model: Claude haiku (classification) Output: CRM activities + PM tasks + drafts + Telegram notify

Related skills

process-analyst — creates the spec
agent-tester — tests the agent
git-workflow — commit and PR

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Gives 0 of the 12 instructions most context ai engineering skills give in 848 tokens

Counted across 1,328 of the 2,349 authors here whose files we hold, read 2026-09-06

Dispatch a fresh subagent for each taskin 76 of 1328, across 59 files
Perform spec compliance review before code quality reviewin 44 of 1328, across 34 files
Dispatch a final code reviewer after all tasksin 38 of 1328, across 26 files
Answer subagent questions before allowing implementationin 36 of 1328, across 26 files
Use the least powerful model capable of the taskin 33 of 1328, across 26 files
Create a TodoWrite list for all tasksin 32 of 1328, across 22 files
Perform a task review after each implementationin 31 of 1328, across 24 files
Extract all tasks and context from the planin 29 of 1328, across 20 files
Provide full task text to subagentsin 28 of 1328, across 20 files
Use git worktrees for isolated workspacesin 25 of 1328, across 20 files
Specify the model explicitly when dispatching a subagentin 23 of 1328, across 18 files
Execute all tasks from the plan without stoppingin 21 of 1328, across 16 files

Said here and by no other author read

Read the reference implementation
Define agent architecture
Use Claude CLI for AI
Use CSV for data
Add logging to stdout and files
Ensure agent idempotency

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.