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 instructions/dubsopenhub/dark-factory/agents-mdgit clone --depth 1 https://github.com/DUBSOpenHub/dark-factoryWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/dubsopenhub/dark-factory/agents-md)<a href="https://agentmods.dev/instructions/dubsopenhub/dark-factory/agents-md"><img src="https://agentmods.dev/badge/instructions/dubsopenhub/dark-factory/agents-md.svg" alt="Measured on agentmods" height="20"></a>What 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 | $0.01617 | $0.01617 |
| Opus 5 | $0.00809 | $0.00809 |
| Sonnet 5 | $0.00323 | $0.00323 |
| Haiku 4.5 | $0.00162 | $0.00162 |
Grade A, and why
dark-factory AGENTS.md 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 5d ago.
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 — 146 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md — Working Guide for AI Agents
This file tells any AI agent how to work effectively on the Dark Factory codebase.
Architecture
Dark Factory is a Copilot CLI skill that orchestrates 8 specialist agents through a gate-driven pipeline with sealed-envelope testing and cross-family model independence (Shadow Score Spec v2.0, Level 4).
SKILL.md (Factory Manager / Orchestrator)
├── agents/product-mgr.md → Phase 1: writes PRD.md
├── agents/qa-sealed.md → Phase 2a: writes sealed tests (×N authors, parallel)
├── agents/architect.md → Phase 2b: writes ARCH.md
├── agents/arch-critic.md → Phase 2.5b: critiques ARCH.md before any code exists
├── agents/lead-eng.md → Phase 3 + 5: writes code, open tests, hardening
├── agents/qa-validator.md → Phase 4: runs both suites, computes Shadow Score
├── agents/red-team.md → Phase 4.5: attacks what the spec forgot
└── agents/outcome-evaluator.md → Phase 7: post-ship outcomes + calibration
Each agent is called via task() with explicit inputs. Agents are stateless — they receive ONLY what the handoff manifest specifies. Phase 5 is the exception: hardening continues the same agent via write_agent, so accumulated context is preserved across cycles.
File Ownership Map
| File/Dir | Owner | Change Rules |
|---|---|---|
SKILL.md |
Orchestrator logic | Most critical file. Changes affect all phases. |
agents/*.md |
Individual agent prompts | Change one agent at a time. Test with a factory run. |
agents/outcome-evaluator.md |
Post-ship analysis prompt | Keep KPI math aligned with PRD template + config outcome_evaluation settings. |
config.yml |
User-tunable settings | Never hardcode values that belong here. Changing roles or families changes conformance — re-run the validator. |
templates/*.md |
Output format templates | Keep in sync with agent prompts that reference them. |
protocols/sealed-envelope.md |
Isolation invariants | Referenced by SKILL.md and agent prompts. |
protocols/model-independence.md |
Why Level 4 exists | The intellectual core. Read before touching config.roles. |
protocols/checkpoint-gate.md |
Autonomy + quality gates | Gate thresholds live in config.gates, not here. |
.github/scripts/validate_conformance.py |
CI conformance check | The only executable in the repo. Keep it dependency-light. |
catalog.yml |
Skill metadata | Update version on every release. Every agents/*.md must be listed. |
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.
- 5d ago First seen · 146 lines · 1,617 tokens per session scan A f3c9ca9ff8bb
dark-factory AGENTS.md is an instructions file published in the GitHub repository DUBSOpenHub/dark-factory (25 stars, last pushed 1mo ago), licensed MIT. It adds 1,617 tokens to every session, about $0.0081 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 2026-08-30.
Other instructions, from other repositories
agent-framework copilot-instructions.md
Copilot instructions for microsoft/agent-framework, covering github copilot instructions, repository structure and architectural decision records (adrs).
agent-framework python.instructions.md
Instructions for microsoft/agent-framework: See AGENTS.md for project structure and package documentation. Detailed conventions are in the agent skills under .github/skills/.
graph-engineering-architectures AGENTS.md
Instructions for Mark393295827/graph-engineering-architectures, covering graph engineering architectures - agent instructions, scope, change contract and required checks.
isomux AGENTS.md
AGENTS.md instructions for nmamano/isomux, covering isomux, how to develop, formatting and linting, key decisions (do not revisit) and project layout.
Starlight-Intelligence-System AGENTS.md
AGENTS.md instructions for frankxai/Starlight-Intelligence-System, covering frank dna, standards, multi-agent systems & agent harnesses, 1. what makes sis unique and 2. the swarm harness process.
paperclip-plugin-company-wizard CLAUDE.md
Instructions for Yesterday-AI/paperclip-plugin-company-wizard, covering claude.md, what this is, commands, architecture and source layout.