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/olyannaa/clickadvisor/agents-mdgit clone --depth 1 https://github.com/olyannaa/clickadvisorWhat 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 | $0.00783 | $0.00783 |
| Opus 5 | $0.00392 | $0.00392 |
| Sonnet 5 | $0.00157 | $0.00157 |
| Haiku 4.5 | $0.00078 | $0.00078 |
Grade A, and why
clickadvisor 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 2d 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 — 87 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Правила Работы Агентов С Репозиторием
Этот файл описывает, как AI-агенты должны работать с ClickAdvisor.
Базовый принцип
ClickAdvisor — local-first анализатор ClickHouse SQL. Доверенный runtime path: детерминированный rule engine, ML evaluation surface и локальный retrieval. Generative LLM не должен становиться источником production-рекомендаций внутри CLI/MCP.
Перед изменениями
- Начинай с чтения
README.md,docs/ARCHITECTURE.md,docs/evaluation.mdи релевантных файлов изdocs/rules/. - Проверяй текущее состояние через
git status -sb. - Не перетирай пользовательские изменения и не используй destructive git commands без явной просьбы.
- Для поиска используй
rg/rg --files.
Правила каталога
- У каждого runtime rule должен быть card в
docs/rules/cards/. - У каждого card со статусом implemented должна быть зарегистрированная
реализация в
clickadvisor/rules/registry.py. - Нельзя добавлять duplicate
rule_id. - Если правилу нужен schema/environment/explain context, отсутствие контекста
должно давать
None, а не догадочную находку. - Tier 2 правила являются advisory. Не добавляй automatic rewrite, если card явно не содержит безопасный шаблон.
- Environment rules должны читать только
QueryContext.environmentи дополнительные доступные context-поля.
Тесты и benchmark
Для нового или измененного правила нужны прямые unit-тесты:
- positive case;
- no-trigger или no-context case;
- проверка ожидаемого
rule_id.
Если правило можно представить в benchmark schema, добавь case в
benchmark/cases/synthetic_expanded/.
Benchmark labels должны отражать реальные multi-label overlaps. В strict mode не должно быть неожиданных FP/FN.
Документация
- Основная документация репозитория ведется на русском языке.
- README должен соответствовать фактическим счетчикам rules/tests/benchmark.
- Не заявляй speedup, production LLM, query replay или автоматические DDL изменения, если этого нет в коде и проверках.
- Enterprise/security формулировки должны подчеркивать zero data egress: SQL, DDL, EXPLAIN и environment context не отправляются во внешние LLM/API.
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.
- 2d ago First seen · 87 lines · 783 tokens per session scan A dc8e40a5619a
clickadvisor AGENTS.md is an instructions file published in the GitHub repository olyannaa/clickadvisor (18 stars, last pushed 9d ago), licensed MIT. It adds 783 tokens to every session, about $0.0039 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
jitsu AGENTS.md
AGENTS.md instructions for jitsucom/jitsu, covering jitsu — agent guidelines, project overview, repository structure, go (/bulker) and node.js / typescript.
openlit AGENTS.md
AGENTS.md instructions for openlit/openlit, covering openlit agent rules, ce/oss boundary, extension points, project hierarchy and ui and security.
Addax AGENTS.md
AGENTS.md instructions for wgzhao/Addax, covering agents.md, 概述, 通用偏好, 编译 and 整体编译.
Tracely-ai CLAUDE.md
Claude Code instructions for Jwuthri/Tracely-ai, covering claude.md, commands, architecture, hard rules and gotchas.
clickhouse-go AGENTS.md
AGENTS.md instructions for ClickHouse/clickhouse-go, a project described as: Golang driver for ClickHouse.
openlit CLAUDE.md
Claude Code instructions for openlit/openlit, a project described as: Open source platform for AI Engineering: OpenTelemetry-native LLM Observability, GPU Monitoring, Guardrails, Evaluations, Prompt Management, Vault, Playground. 🚀💻 Integrates with 50+ LLM Providers, VectorDBs, Agent Frameworks and GPUs.