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/syndicalt/zaxy/agents-mdgit clone --depth 1 https://github.com/syndicalt/zaxyWrote 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/syndicalt/zaxy/agents-md)<a href="https://agentmods.dev/instructions/syndicalt/zaxy/agents-md"><img src="https://agentmods.dev/badge/instructions/syndicalt/zaxy/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 | $0.01417 | $0.01417 |
| Opus 5 | $0.00709 | $0.00709 |
| Sonnet 5 | $0.00283 | $0.00283 |
| Haiku 4.5 | $0.00142 | $0.00142 |
Grade A, and why
zaxy 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 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 — 101 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md — working on Zaxy
Zaxy is an event-sourced memory runtime for agent fleets: an append-only,
hash-chained JSONL log (Eventloom) is the source of truth; an embedded
LadybugDB graph projection and every cache are replayable derivatives; recall
is a cited Memory Checkout. Published to PyPI as zaxy-memory (CLI
binary zaxy; the bare zaxy name on PyPI is an unrelated package). Docs
site: https://zaxy.io.
Controlling docs: CLAUDE.md (operating manual: conventions, named
failure modes, quality bars, escalation rules) + docs/architecture.md +
README.md. This file is the condensed cross-tool orientation; when it and
the code disagree, the code wins — fix this file.
Architecture in five invariants
.eventloom/*.jsonlis append-only truth. Never edit, truncate, or delete it. Everything under.eventloom/projections/is derived and rebuildable — move it aside (neverrm) and replay rebuilds it.- Embedded LadybugDB is the default backend: file-based, in-process, zero-daemon. Neo4j is an optional sidecar; pgGraph and LatticeDB are experimental. Nothing may assume Neo4j is primary.
- One owner per store (exclusive write lock); concurrent owners corrupt the
WAL. Owner locking, reaping, self-heal, and the pre-open bloat guard live
in the embedded runtime — see
docs/runbook.md. - MCP is the primary interface (
memory_checkoutis the front door); the CLI is the human/debug surface. The MCP tool contract is snapshot-pinned indocs/examples/mcp-tool-contract.json— changing a tool means regenerating the snapshot in the same PR. - Recall is cited or it doesn't count: current facts carry
eventloom://citations, and agents are told to trust only cited facts.
Decision record (condensed)
| ADR | Decision | Status |
|---|---|---|
| 1 | Event-sourced JSONL over mutable state | active |
| 2 | Hybrid extraction: deterministic rules first, LLM fallback | active |
| 3 | Graph projection backend | superseded: embedded LadybugDB default; Neo4j sidecar opt-in |
| 4 | Hybrid retrieval (exact + BM25 + graph traversal + embeddings) | active |
| 5 | Pathlight observability integration | active, optional |
| 6 | MCP as the primary agent interface | active |
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 · 101 lines · 1,417 tokens per session scan A c220d89d768b
zaxy AGENTS.md is an instructions file published in the GitHub repository syndicalt/zaxy (17 stars, last pushed 24d ago), licensed MIT. It adds 1,417 tokens to every session, about $0.0071 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
hindsight CLAUDE.md
Claude Code instructions for vectorize-io/hindsight, covering claude.md, project overview, development commands, local development (api + ui) and start both api server and control plane ui.
honcho CLAUDE.md
Claude Code instructions for plastic-labs/honcho, covering claude.md, honcho overview, what is honcho?, core concepts and peer paradigm.
Dragon-Brain CLAUDE.md
Instructions for iikarus/Dragon-Brain, covering dragon brain — claude.md, the harness, audit remediation (april–may 2026, complete 2026-05-09), the lie this audit closed and the contract that matters now.
dense-mem AGENTS.md
Instructions for markhuangai/dense-mem, covering dense-mem repository guidance, project context, architecture decision records, current stack and target architecture.
EGC copilot-instructions.md
Copilot instructions for Fmarzochi/EGC, covering copilot instructions for egc, repository conventions, readme and translations, code review priorities and egc project memory.
brainlayer AGENTS.md
Instructions for EtanHey/brainlayer, covering brainlayer — a letter from etan, brainlayer agent notes (operational), review guidelines, key paths and database.