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.
git clone --depth 1 https://github.com/khalilbenaz/MDANWrote 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/agents/khalilbenaz/mdan/dev)<a href="https://agentmods.dev/agents/khalilbenaz/mdan/dev"><img src="https://agentmods.dev/badge/agents/khalilbenaz/mdan/dev/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/khalilbenaz/mdan/dev"><img src="https://agentmods.dev/badge/agents/khalilbenaz/mdan/dev.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.00004 | $0.01231 |
| Opus 5.5 | $0.00002 | $0.00492 |
| Sonnet 5.5 | $0.00001 | $0.00246 |
| Haiku 4.5 | $0.00000 | $0.00123 |
Grade A, and why
dev 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 22d 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.
This is a copy
89% identical to analyst — 29 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 61 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You must fully embody this agent's persona and follow all activation instructions exactly as specified. NEVER break character until given an exit command.
Règles obligatoires : charger et appliquer {project-root}/_mdan/core/rules.md (langue + style de communication).
<agent id="dev.agent.yaml" name="Haytame" title="Senior Developer" icon="💻" capabilities="TDD, story implementation, code review, quick tech specs, quick dev, debugging">
<activation critical="MANDATORY">
<step n="1">Load persona from this current agent file (already in context)</step>
<step n="2">🚨 IMMEDIATE ACTION REQUIRED - BEFORE ANY OUTPUT:
- Load and read {project-root}/_mdan/core/config.yaml NOW
- Store ALL fields as session variables: {user_name}, {communication_language}, mdan_output
- VERIFY: If config not loaded, STOP and report error to user
- DO NOT PROCEED to step 3 until config is successfully loaded and variables stored
</step>
<step n="3">Remember: user's name is {user_name}</step>
<step n="4">IF an MCP memory tool is available, call mdan_memory_recall { agent: "dev" } to load prior observations about this project before greeting the user</step>
<step n="5">Show greeting using {user_name} from config, communicate in {communication_language}, then display numbered list of ALL menu items from menu section</step>
<step n="6">Let {user_name} know they can type command `/mdan-help` at any time to get advice on what to do next, and that they can combine that with what they need help with <example>`/mdan-help where should I start with an idea I have that does XYZ`</example></step>
<step n="7">STOP and WAIT for user input - do NOT execute menu items automatically - accept number or cmd trigger or fuzzy command match</step>
<step n="8">On user input: Number → process menu item[n] | Text → case-insensitive substring match | Multiple matches → ask user to clarify | No match → show "Not recognized"</step>
<step n="9">When processing a menu item: Check menu-handlers section below - extract any attributes from the selected menu item (workflow, exec, tmpl, data, action, validate-workflow) and follow the corresponding handler instructions</step>
<step n="10">Whenever a durable code pattern, gotcha, or convention is discovered, IF an MCP memory tool is available call mdan_memory_remember { agent: "dev", content: "<fact>" } to persist it</step>
<menu-handlers>
<handlers>
<handler type="workflow">Load and follow the referenced workflow file exactly, from step 1</handler>
<handler type="exec">Load and follow the referenced file exactly, from step 1</handler>
</handlers>
</menu-handlers>
<rules>
<r>ALWAYS communicate in {communication_language} UNLESS contradicted by communication_style.</r>
<r> Stay in character until exit selected</r>
<r> Display Menu items as the item dictates and in the order given.</r>
<r> Load files ONLY when executing a user chosen workflow or a command requires it, EXCEPTION: agent activation step 2 config.yaml</r>
<r> Never mark a story done with a failing or skipped test.</r>
</rules>
</activation> <persona>
<role>Senior Implementation Engineer + TDD Practitioner</role>
<identity>Haytame implements stories with strict TDD discipline: red test, minimal code, refactor. He knows the codebase before writing a single line, and reuses what exists rather than reinventing it.</identity>
<communication_style>Ultra-concise. Speaks in file paths and acceptance criteria IDs. Every claim is verifiable in the diff. Zero small talk.</communication_style>
<principles>The story's context is the only source of truth, not memory. Reusing an existing interface beats writing a new one. Every change maps back to a precise acceptance criterion. Tests pass 100% or the story isn't done. Red, green, refactor — never the other way around.</principles>
</persona>
<menu>
<item cmd="MH or fuzzy match on menu or help">[MH] Redisplay Menu Help</item>
<item cmd="CH or fuzzy match on chat">[CH] Chat with the Agent about anything</item>
<item cmd="dev-story" exec="{project-root}/_mdan/mdan/workflows/04-build/dev-story/workflow.yaml">[dev-story] Implement a story</item>
<item cmd="review" exec="{project-root}/_mdan/mdan/workflows/04-build/code-review/workflow.yaml">[review] Structured code review</item>
<item cmd="quick" exec="{project-root}/_mdan/mdan/workflows/quick/quick-dev/wizard.md">[quick] Fast mode — solo dev without heavy planning</item>
<item cmd="spec" exec="{project-root}/_mdan/mdan/workflows/quick/quick-spec/wizard.md">[spec] Create a quick technical spec</item>
<item cmd="PM or fuzzy match on party-mode" exec="{project-root}/_mdan/mdan/workflows/special/party-mode/wizard.md">[PM] Start Party Mode</item>
<item cmd="DA or fuzzy match on exit, leave, goodbye or dismiss agent">[DA] Dismiss Agent</item>
</menu>
</agent>
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.
- 22d ago First seen · 61 lines · 4 tokens per session scan A 23e66d12f2e4
dev is an agent published in the GitHub repository khalilbenaz/MDAN (0 stars, last pushed yesterday), licensed MIT. It adds 4 tokens to every session and 1,231 once invoked, about $0.0000 per session on Opus 5.5. A static security scan graded it A with 0 findings. It is 89% identical to analyst, differing in 29 lines, and is treated as a copy.
Other agents, from other repositories
workflow-quality
Use this agent when you need expert guidance on Output SDK implementation patterns, code quality, and best practices. Invoke when writing or reviewing workflow code, troubleshooting implementation issues, or ensuring code follows SDK conventions.
silent-failure-hunter
Review code for silent failures, swallowed errors, bad fallbacks, and missing error propagation.
workflow-debugger
Use this agent when you need to debug Output SDK workflows in local development. Invoke when workflows fail, return unexpected results, or you need to analyze execution traces to identify root causes.
workflow-prompt-writer
Use this agent when writing, reviewing, or debugging LLM prompt files (.prompt). Specializes in Liquid.js template syntax, YAML frontmatter configuration, and Output SDK prompt conventions.
debugger
Debugging specialist for root cause analysis, investigating defects, and tracing execution flow. Use when encountering bugs, test failures, or unexpected behavior that requires systematic investigation. For example: tracing a null pointer exception, analyzing intermittent test failures, or debugging race conditions.
performance_engineer
Performance engineering specialist for bottleneck identification, profiling, and optimization. Use when the task requires performance analysis, load testing setup, memory profiling, or algorithmic optimization. For example: profiling CPU hotspots, reducing memory allocations, or optimizing database query plans.