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/kdowding/stela/claude-mdgit clone --depth 1 https://github.com/kdowding/StelaWrote 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/kdowding/stela/claude-md)<a href="https://agentmods.dev/instructions/kdowding/stela/claude-md"><img src="https://agentmods.dev/badge/instructions/kdowding/stela/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 | $0.01002 | $0.01002 |
| Opus 5 | $0.00501 | $0.00501 |
| Sonnet 5 | $0.00200 | $0.00200 |
| Haiku 4.5 | $0.00100 | $0.00100 |
Grade A, and why
Stela 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 — 37 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Stela
Self-hosted platform to host and share agent-built artifacts (self-contained HTML — the shape Claude's artifacts popularized). Any MCP-capable agent can publish; share privately, server-wide, or with named people; every artifact gets a stable URL, immutable revision history, and pinned visual comments. Stela: an inscribed stone slab erected for public display.
Stack
pnpm workspaces, TypeScript strict end-to-end, SvelteKit + Svelte 5 (runes, adapter-node).
packages/shared— Zod schemas/types (Artifact, Version, Comment, Anchor, sharing DTOs) imported by UI, API, and MCP.packages/app— the deployable: UI + API + the remote MCP endpoint (/mcp, Streamable HTTP) + Stela's own OAuth 2.1 AS (PKCE, DCR) for MCP clients.packages/mcp— stdio MCP server for CLI agents (stela-mcp): publish by file path, auto-versioning via~/.stela/artifacts.json.
Identity & storage (the two pluggable seams)
- Auth = trusted headers. An identity-aware proxy in front of the app injects identity; Stela has NO account system — first sight of a new id is enrollment.
AUTH_MODE=header+AUTH_HEADER_ID/NAME/EMAIL(generic), orAUTH_PRESET=easyauth(Azure Easy Auth contract). Dev builds use a dev-shim user. Prod fails loud at boot if auth is unconfigured. OptionalAUTH_LOGIN_URLredirects anonymous browser navs. - Storage =
Storeinterface (lib/server/storage/types.ts) with two drivers:sqlite(default —node:sqlite, oneDATA_DIR/stela.db, WAL, STRICT tables) andazure(Tables + Blobs, ETag concurrency).STORAGE_DRIVERselects; prod requires it explicitly, sqlite requires a writableDATA_DIR. Driver-agnostic semantics (title sync, republish dedup, hashing, PKCE, TTLs, token cap) live instorage/shared.ts.
Hard rules
- Never touch a storage SDK (
node:sqlite,@azure/*) outsidepackages/app/src/lib/server/storage/— depend ongetStore()/ theStoreinterface. - Never read identity headers outside
packages/app/src/lib/server/auth/— usegetCurrentUser()/authenticateApiKey()/locals.user. - Route handlers go through
lib/server/guards.ts(auth/authz/validation, same-origin CSRF for browser mutations). - Artifacts are immutable per version; republish appends (identical bytes dedup); the share URL stays stable. Comments are scoped to the version they were made on.
- Artifacts render in an opaque-origin sandboxed iframe under
default-src 'none'CSP — no network egress.fetchRemoteHtml.tsis the app's only outbound request path and is SSRF-hardened; keep it that way. svelte-check+tsc --strictmust pass with 0 errors; tests vianpx pnpm -r --if-present test(vitest).
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 · 37 lines · 1,002 tokens per session scan A a1f9e43de2dd
Stela CLAUDE.md is an instructions file published in the GitHub repository kdowding/Stela (0 stars, last pushed 1mo ago), licensed MIT. It adds 1,002 tokens to every session, about $0.0050 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
rekal AGENTS.md
Instructions for janbjorge/rekal, covering rekal: agent instructions, what this project is, architecture, key rules and re-expansion bar.
Vault-Agent-Memory AGENTS.md
Instructions for zycaskevin/Vault-Agent-Memory, covering agent instructions for vault agent memory, common install architecture, second decision: optional features, first decision: database scope and safe agent workflow.
rekal CLAUDE.md
Instructions for janbjorge/rekal, a project described as: Long-term memory for LLMs. MCP server backed by hybrid search in a single SQLite file.
zen CLAUDE.md
Claude Code instructions for sheshbabu/zen, covering claude.md, development commands, build commands, dependencies and go build tags.
db-mcp copilot-instructions.md
Copilot instructions for neverinfamous/db-mcp, covering db-mcp — copilot code review context, project overview, session context, coding standards and naming.
artel CLAUDE.md
Claude Code instructions for NicolasPrimeau/artel, covering artel, what it is, stack, layout and core primitives.