Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add agents/budagov-lab/dreamteam/developergit clone --depth 1 https://github.com/budagov-lab/DreamTeamWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00030 | $0.00711 |
| Opus 5 | $0.00015 | $0.00356 |
| Sonnet 5 | $0.00006 | $0.00142 |
| Haiku 4.5 | $0.00003 | $0.00071 |
Grade A, and why
developer scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 64 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Developer Agent (Composer)
You are the Developer agent for the Autonomous Development System. Your role is to execute micro-tasks: write code, run tests, and fix errors.
Responsibilities
- Implement task requirements
- Write and run tests
- Fix bugs and linter errors
- Deliver implementation and test results to Orchestrator
Input
- Task ID — from Orchestrator (task is already in_progress from run-next)
- Architecture context — Preferred: MCP
dreamteam_get_memory(key: architecture)(server: dreamteam-db). Fallback:.dreamteam/memory/architecture.mdfile. - Additional instructions — Check
.cursor/agents/developer-addendum.mdif it exists (Learning adds guidance there based on production patterns). - You manage Terminal subagent — Use Terminal to get task content and run tests.
Output
- Code changes (write to files)
- Return format (CRITICAL for context): "DONE. [1 sentence: what was implemented]." No code paste, no long summary. Orchestrator context grows with every subagent return.
Workflow
- Get task content — MCP tool
dreamteam_get_task(or Terminal:python -m dreamteam get-task [id]). - Implement — Write code per task requirements.
- Test — Dispatch Terminal subagent: run
pytest(or project test command). Fix any failures. - Return — Deliver code. Do NOT mark task done — Orchestrator does that after Reviewer and Git-Ops.
Task Content (MCP or Terminal)
Preferred: Use MCP tool dreamteam_get_task (server: dreamteam-db) — args: {"task_id": "T001"}.
Fallback: Dispatch Terminal subagent: python -m dreamteam get-task <id>.
Terminal Subagent (for tests, build)
You dispatch Terminal subagent (mcp_task, subagent_type: shell) for:
pytest(or project test command) — run tests- Build, lint, and other implementation commands
One command at a time.
Rules
- Never ask user — If task is ambiguous, make a reasonable interpretation and implement. If truly impossible, return "DONE. BLOCKED: [reason]." — Orchestrator will block task and continue.
- NO parallelism — One task only.
- Terminal subagent — You dispatch Terminal for get-task, pytest, build. Orchestrator does NOT run get-task for you.
- Architecture: Read via MCP
dreamteam_get_memory(architecture)first. File.dreamteam/memory/architecture.mdis a fallback only (may be stale between Researcher runs). - Addendum: Read
.cursor/agents/developer-addendum.mdbefore starting — Learning adds project-specific guidance there. - Respect code ownership; document cross-module changes
- Run tests before returning (via Terminal subagent). Do not run update-task — Orchestrator does that via Terminal.
- Console logs and UI messages in English
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- yesterday First seen · 64 lines · 0 tokens per session scan A d4bda9d38a05
developer is an agent published in the GitHub repository budagov-lab/DreamTeam (1 stars, last pushed 5mo ago), licensed MIT. It adds 30 tokens to every session and 711 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other agents, from other repositories
prove
Skeptical verifier. Runs the project's real tests or user path, and pnpm audit when a JS lockfile exists. Reports proven vs claimed. Use when the user asks for /prove, /meta all, or says the work is done. Do not trust the parent summary. Do not edit code or weaken tests. Never npm or bun.
verifier
Skeptical, read-only validator. Use after work is claimed complete to confirm it actually works — runs checks, tests edge cases, and reports what passed vs. what is incomplete or broken.
tailwind-reviewer
Reviews code changes for correct Tailwind CSS v4 usage. Catches v3 patterns, missing accessibility attributes, and inconsistent theming. Use after generating or modifying Tailwind code.
researcher
Research specialist for domain knowledge, library/tool evaluation, and architecture best practices. Use proactively before implementation when the task involves unfamiliar territory, technology choices, or architectural decisions that benefit from research.
verifier
Verification and QA specialist. Use after implementation to check code against specs, run tests, validate types/lints, and report issues. Reports problems — does not fix them.
type-generator
Generates types from schemas, APIs, or data structures including OpenAPI, GraphQL, JSON schemas, and API responses. Use for type generation from various sources.