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/nimblebraininc/synapse-research/claude-mdgit clone --depth 1 https://github.com/NimbleBrainInc/synapse-researchWrote 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/nimblebraininc/synapse-research/claude-md)<a href="https://agentmods.dev/instructions/nimblebraininc/synapse-research/claude-md"><img src="https://agentmods.dev/badge/instructions/nimblebraininc/synapse-research/claude-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 | $0.02353 | $0.02353 |
| Opus 5 | $0.01177 | $0.01177 |
| Sonnet 5 | $0.00471 | $0.00471 |
| Haiku 4.5 | $0.00235 | $0.00235 |
Grade C, and why
synapse-research CLAUDE.md scanned grade C with 1 finding 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 4d 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.
Recursive force deletehighDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
rm -rf deps/ && uv pip install --target ./deps . How it starts
The opening of the file, as written. The whole thing — 160 lines — stays where its author put it; the contents beside it link to each section on GitHub.
synapse-research
Deep-research MCP app with a task-augmented tool that runs GPT-Researcher under the hood. Two-project architecture: Python MCP server (FastMCP) + React/Vite UI (Synapse SDK).
Architecture
synapse-research/
├── src/mcp_research/ Python MCP server (FastMCP)
│ ├── server.py Tool registration, orphan reaper, ui:// resource
│ └── worker.py run_research + _MCPLogHandler (GPT-Researcher wrapper)
├── schemas/ Upjack entity JSON schemas
│ └── research_run.schema.json
├── skills/ Natural-language skill definitions bundled with the app
│ └── research-runs/SKILL.md
├── ui/ React + Vite project (Synapse SDK)
│ ├── src/App.tsx Live dashboard of runs via useDataSync
│ ├── vite.config.ts synapseVite() plugin for dev
│ └── dist/index.html Built single-file bundle
└── tests/ pytest-asyncio suite — keyless (FakeGPTR fixture)
How it connects: start_research is registered as a task-augmented MCP tool (execution.taskSupport: "optional"). The worker wraps GPT-Researcher and streams progress through two channels in lockstep — ctx.report_progress for the MCP task status stream (drives the chat UI), and app.update_entity for the Upjack entity store (drives the Synapse sidebar dashboard).
Commands
# Install server deps
uv sync
# Install UI deps + build
cd ui && npm install && npm run build && cd ..
# Run tests (keyless — no API keys required)
uv run pytest tests/ -v
# Lint + typecheck + format check
uv run ruff check src/ tests/
uv run ruff format --check src/ tests/
uv run ty check src/
# Run stdio (Claude Desktop, mpak, any MCP client)
uv run python -m mcp_research.server
# Run HTTP (NimbleBrain platform)
uv run uvicorn mcp_research.server:app --port 8002
# Build deps/ for mpak bundling (regenerates 460+ vendored packages)
rm -rf deps/ && uv pip install --target ./deps .
# Bump version across manifest.json, pyproject.toml, __init__.py
make bump VERSION=0.2.0
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.
- 4d ago First seen · 160 lines · 2,353 tokens per session scan C cb08d05e5721
synapse-research CLAUDE.md is an instructions file published in the GitHub repository NimbleBrainInc/synapse-research (0 stars, last pushed 3mo ago), licensed MIT. It adds 2,353 tokens to every session, about $0.0118 per session on Opus 5. A static security scan graded it C with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
primr AGENTS.md
AGENTS.md instructions for blisspixel/primr, covering agents.md - primr, agent-host default, when this is the right tool, before first invocation and detecting mcp vs cli.
primr CLAUDE.md
Claude Code instructions for blisspixel/primr, covering claude.md - the contract for building primr, quick start, language and runtime choices, architecture pointers and use the one seam - don't invent a sixth way.
Autosearch CLAUDE.md
Instructions for 0xmariowu/Autosearch, covering autosearch — project rules, what this is, local test (run before push), session checklist and workflow.
Autosearch copilot-instructions.md
Instructions for 0xmariowu/Autosearch, covering copilot instructions for agentlint, hard rules — never violate, scope discipline for this repo, commit and pr format and ci-fix specific rules (ci-doctor / @copilot fix ci).
blockrun-mcp AGENTS.md
AGENTS.md instructions for BlockRunAI/blockrun-mcp, covering blockrun mcp, commands, project structure, key dependencies and install in codex.
openrouter-mcp-multimodal AGENTS.md
AGENTS.md instructions for stabgan/openrouter-mcp-multimodal, covering agent instructions, before you ship, releasing (read this before publishing), short version and version files (must all match package.json).