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/refactornator/nanoflash/claude-mdgit clone --depth 1 https://github.com/refactornator/nanoflashWrote 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/refactornator/nanoflash/claude-md)<a href="https://agentmods.dev/instructions/refactornator/nanoflash/claude-md"><img src="https://agentmods.dev/badge/instructions/refactornator/nanoflash/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 | $0.02775 | $0.02775 |
| Opus 5 | $0.01388 | $0.01388 |
| Sonnet 5 | $0.00555 | $0.00555 |
| Haiku 4.5 | $0.00278 | $0.00278 |
Grade A, and why
nanoflash CLAUDE.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 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.
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 — 261 lines — stays where its author put it; the contents beside it link to each section on GitHub.
NanoFlash
Personal Gemini agent harness. See README.md for philosophy and setup.
Quick Context
Single Node.js process with skill-based channel system. Channels (WhatsApp, Telegram, Slack, Discord, Gmail) are skills that self-register at startup. Messages route to a Gemini agent loop running in containers (Linux VMs). Each group has isolated filesystem and memory.
Architecture
User message → Channel (Telegram/WhatsApp/etc.)
→ SQLite (store message)
→ Polling loop (src/index.ts, every 2s)
→ container-runner.ts spawns Apple Container / Docker
→ agent-runner (container/agent-runner/src/index.ts)
→ Gemini API with tool loop
→ stdout: NANOFLASH_OUTPUT_START/END markers
→ Host parses output, sends response via channel
→ IPC files for side-effects (messages, reactions, tasks)
Two-process boundary: The host process (src/) manages channels, DB, IPC, and container lifecycle. The agent process (container/agent-runner/) runs inside a Linux container with only mounted directories visible. These communicate via stdin/stdout (initial prompt + output markers) and filesystem IPC (/workspace/ipc/).
Key Files
| File | Purpose |
|---|---|
src/index.ts |
Orchestrator: state, message loop, agent invocation |
src/container-runner.ts |
Spawns agent containers, builds mounts and env vars, parses output markers |
src/channels/registry.ts |
Channel registry (self-registration at startup) |
src/ipc.ts |
IPC watcher: processes message/reaction/task files from containers |
src/router.ts |
Formats messages as XML for the agent, strips <internal> tags on output |
src/config.ts |
All configuration: trigger pattern, paths, intervals, Gemini models |
src/gemini-media.ts |
Host-side media analysis — images inline, video/audio via Gemini File API (up to 2 GB) |
src/task-scheduler.ts |
Runs scheduled tasks on cron/interval/once schedules |
src/db.ts |
SQLite operations (messages, sessions, groups, tasks) |
src/types.ts |
Channel interface, RegisteredGroup, NewMessage, etc. |
groups/{name}/CLAUDE.md |
Per-group system instruction (loaded as Gemini systemInstruction) |
container/agent-runner/src/index.ts |
The agent — Gemini tool loop, all tool implementations |
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 · 261 lines · 2,775 tokens per session scan A ec268ee3e0ed
nanoflash CLAUDE.md is an instructions file published in the GitHub repository refactornator/nanoflash (5 stars, last pushed 4mo ago), licensed MIT. It adds 2,775 tokens to every session, about $0.0139 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 instructions, from other repositories
appwrite AGENTS.md
AGENTS.md instructions for appwrite/appwrite, covering appwrite, commands, stack, layout and libraries.
selfhost-ai CLAUDE.md
Claude Code instructions for kossakovsky/selfhost-ai, covering claude.md, project overview, core architecture, key files and installation flow.
openlakeforge AGENTS.md
Instructions for malon64/openlakeforge, covering agent and contributor guide, what this project is, orientation — read in this order, repository map and architectural rules.
docker-zoneminder CLAUDE.md
Instructions for jantman/docker-zoneminder, covering claude.md, project overview, build and test, ci/cd and architecture.
mitos CLAUDE.md
Instructions for mitos-run/mitos, covering claude.md, project overview, operating principles, commands and architecture.
MoaV AGENTS.md
AGENTS.md instructions for MotherofallVPNs/MoaV, covering what moav is, install & run, everyday operations, dashboards & access — what to hand the owner and where things live.