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/0xryanlee/salva/claude-mdgit clone --depth 1 https://github.com/0xRyanlee/salvaWrote 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/0xryanlee/salva/claude-md)<a href="https://agentmods.dev/instructions/0xryanlee/salva/claude-md"><img src="https://agentmods.dev/badge/instructions/0xryanlee/salva/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 | $0.02593 | $0.02593 |
| Opus 5 | $0.01296 | $0.01296 |
| Sonnet 5 | $0.00519 | $0.00519 |
| Haiku 4.5 | $0.00259 | $0.00259 |
Grade A, and why
salva 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 — 192 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Salva Runtime — Project Context for Claude Code
What This Project Is
Salva Runtime is a self-hosted Discovery Intelligence Runtime — a structured retrieval and entity extraction service designed to be called by AI agents, CLI tools, and LLMs via stable APIs or MCP.
It is not a scraper. It is not a UI product. It is a composable intelligence pipeline that accepts structured intent and returns scored entities, evidence chains, relations, and telemetry.
Core Product Philosophy
Event-Triggered, Not Schedule-Driven
Salva's core business logic is event-triggered close-loop execution:
trigger (agent / CLI / API call)
→ structured intent in
→ multi-round retrieval + processing
→ scored entities + evidence out
→ caller acts on result
This is a deliberate choice. Salva does NOT own scheduling, polling, or monitoring loops. Those concerns belong to the caller (agent, CLI, or orchestrator). Salva responds to calls — it does not initiate them.
If a caller wants recurring execution, they call Salva on a cron from their own scheduler. Salva stays stateless between calls.
API-First, Agent-Native
The primary integration surface is:
- REST API (
/v1/discover,/v1/jobs,/v1/runs) — direct HTTP from any caller - MCP Server — for Claude Code, Claude Desktop, and any MCP-compatible agent
- CLI skill wrapper — for terminal-native agent workflows and Codex integration
- Python SDK — thin wrapper for Python-based agent code
There is no GUI in this repository. A GUI product may be developed as a separate fork — see "GUI Fork" section below.
Deterministic Pipeline First
LLMs are bounded reasoning modules, not the pipeline itself. The enrichment order is:
keyword expansion (deterministic)
→ retrieval (multi-provider, policy-aware)
→ optional bounded LLM query-proposal step (scoped prompt, proposes
follow-up queries only — not free-form reasoning; amended 2026-07-21,
see below)
→ extract → normalize → dedupe → classify → score (all deterministic)
→ enrichment (LLM via omlx — scoped, bounded prompts only)
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 · 192 lines · 2,593 tokens per session scan A 3d4a39370833
salva CLAUDE.md is an instructions file published in the GitHub repository 0xRyanlee/salva (2 stars, last pushed 26d ago), licensed Apache-2.0. It adds 2,593 tokens to every session, about $0.0130 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
fastapi-langgraph-agent-production-ready-template AGENTS.md
Instructions for wassim249/fastapi-langgraph-agent-production-ready-template, covering ai agent development guide, quick commands, project structure, project overview and quick reference: critical rules.
AgentCore AGENTS.md
Instructions for Lawofall/AgentCore, covering agents.md, 先读哪里, docs 与 rules and 开发 / 测试（最短）.
Vibelution AGENTS.md
AGENTS.md instructions for CCDawn/Vibelution, covering vibelution agent rules, 1. identity and priority, 2. global red lines, 3. start and routing and 3.0 每任务先 router.
akb AGENTS.md
AGENTS.md instructions for dnotitia/akb, covering akb project guide, architecture, frontend design system, 2-layer mcp architecture and deployment.
akb CLAUDE.md
Claude Code instructions for dnotitia/akb, covering claude guide for akb, mission, default work loop, pm collaboration and docs state machine.
guidepost-mcp CLAUDE.md
Claude Code instructions for shogo-hs/guidepost-mcp, covering claude.md — guidepost-mcp, コマンド, モジュール, 樹形図のスキーマ and 規約.