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/mclaut/ariadne/agents-mdgit clone --depth 1 https://github.com/mclaut/ariadneWhat 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 | $0.01052 | $0.01052 |
| Opus 5 | $0.00526 | $0.00526 |
| Sonnet 5 | $0.00210 | $0.00210 |
| Haiku 4.5 | $0.00105 | $0.00105 |
Grade A, and why
ariadne 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 3d 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.
Copies of this mod
1 near-identical copy found in the catalogue:
- ariadne CLAUDE.md — 94% identical, 10 lines differ
How it starts
The opening of the file, as written. The whole thing — 71 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md — Ariadne
What this is
Ariadne is a local-first, multilingual memory server for Codex (and any MCP client): Go + Qdrant + bge-m3, hybrid dense+sparse search. It replaces embedded vector-DB memory backends that crash or lock under several concurrent sessions — Qdrant is a server, so the single-writer/lock-starvation class is gone. Public repo.
Build / test / lint
- Build everything:
go build ./... - Install a runtime binary:
go build -o ~/.ariadne/bin/<name> ./cmd/<name>(ariadne,ariadnectl,ariadne-tray,import,ariadne-hook,install) - Lint (must be clean before every commit):
golangci-lint run— config.golangci.yml - Format:
golangci-lint fmt - The
poc/experiments are a separate module:cd poc && go build .
Architecture
internal/store— storage core: bge-m3 dense embedding (Ollama) + BM25 sparse (pure Go tokenizer; Qdrant computes IDF) fused with RRF server-side. No MCP concerns.cmd/ariadne— MCP server (stdio). Tools:memory_save,memory_recall(semantic by query with required wing, explicit all-wing audit opt-in, or exact by id; optional room/collection scopes),memory_delete,memory_move(curate: delete by id, re-home/re-tag by id).cmd/ariadnectl— control/health core;cmd/ariadne-tray(localized system-tray monitor, macOS/Linux/Windows) is a thin viewer over it. The CLI owns status, metrics, service control, backup/restore/export, maintenance/consolidation, secret quarantine and pending-approval inspection.cmd/import— backfill (chromadb sqlite / markdown memfiles / JSONL);-synckeeps memfile chunks true to disk.cmd/hook(ariadne-hook) — Codex session hooks: SessionStart auto-recall, SessionEnd curated auto-capture (local chat model → one diary memory, never raw).cmd/eval— deterministic ranking regressions plus judged ranked-run comparison (Recall/MRR/nDCG); it does not contact Qdrant, Ollama or remote services.cmd/install— one-shot installer for macOS/Linux;install.ps1supplies the Windows path. Both preserve/reuse existing infrastructure where possible.internal/approval,internal/metrics,internal/activity,internal/maintenance,internal/qdrantauth,internal/retrievaleval, andinternal/secretguard— shared approval, observability, retry, transport, evaluation, operation-journal, and credential-detection primitives.
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.
- 3d ago First seen · 71 lines · 1,052 tokens per session scan A 488f2a290fa0
ariadne AGENTS.md is an instructions file published in the GitHub repository mclaut/ariadne (3 stars, last pushed 10d ago), licensed MIT. It adds 1,052 tokens to every session, about $0.0053 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
coco-search CLAUDE.md
Instructions for VioletCranberry/coco-search, covering project instructions, project overview, tool routing (mandatory), development setup and prerequisites: docker, uv (python package manager).
ClawMem AGENTS.md
Instructions for yoloshii/ClawMem, covering clawmem — agent quick reference, inference at a glance, install, or step by step and memory retrieval (90/10 rule).
ClawMem CLAUDE.md
Instructions for yoloshii/ClawMem: Claude Code reads CLAUDE.md natively (not AGENTS.md), so this file imports the root SSOT.
Connapse CLAUDE.md
Instructions for Destrayon/Connapse, covering claude.md, build & run, build, run web app (requires backing services) and full stack via docker.
OpenNotebookLM AGENTS.md
Instructions for tom1030507/OpenNotebookLM, covering repository guidelines, project structure & module organization, build, test, and development commands, coding style & naming conventions and testing guidelines.
subcog CLAUDE.md
Instructions for zircote/subcog, covering claude.md, project overview, key capabilities, build commands and primary development workflow.