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/azurecosmosdb/cosmosdb-agent-kit/code-reviewgit clone --depth 1 https://github.com/AzureCosmosDB/cosmosdb-agent-kitWhat 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 | $0.00860 | $0.00860 |
| Opus 5 | $0.00430 | $0.00430 |
| Sonnet 5 | $0.00172 | $0.00172 |
| Haiku 4.5 | $0.00086 | $0.00086 |
Grade A, and why
cosmosdb-agent-kit code-review.instructions.md scanned grade A with 0 findings against 26 rules in 11 categories โ prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency โ 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.
How it starts
The opening of the file, as written. The whole thing โ 75 lines โ stays where its author put it; the contents beside it link to each section on GitHub.
When performing a code review, apply these checks for the cosmosdb-agent-kit repository.
The full checklist, field requirements, and per-section details live in
.github/skills/code-review/checklist.md
โ read that file for the complete rules. This file summarizes the key gates.
Severity Tiers
- ๐ด Blocking: Must fix before merge.
- ๐ก Recommendation: Should fix.
- ๐ข Suggestion: Nice to have.
Key Gates (๐ด Blocking)
Rule files (skills/*/rules/*.md):
- Frontmatter:
title,impact(CRITICAL|HIGH|MEDIUM-HIGH|MEDIUM|LOW-MEDIUM|LOW),impactDescription,tags - Body:
**Incorrect+**Correctsections with fenced code blocks - Filename:
{prefix}-{description}.md(model-, partition-, query-, sdk-, index-, throughput-, global-, monitoring-, pattern-, tooling-, vector-)
Build:
AGENTS.mdis generated on demand (npm run build) and is not committed; do not require it in rule PRs
Eval tasks (evals/**/*.yaml):
- Required fields:
id,name,description,tags,inputs.prompt,expected.outcomes
Test scenarios (testing-v2/scenarios/):
api-contract.yaml: camelCase fields,health:section requiredtests/conftest.py: deterministic data only โ no uuid4(), random, faker
Scripts: Changes to compile.js/validate.js must not break existing rules.
Workflows: No hardcoded secrets. Use ${{ secrets.* }} or ${{ github.token }}.
General: No secrets, API keys, or connection strings anywhere.
Rule Provenance (๐ด Blocking)
Rules in this repo are created by the automated evaluation loop (LLM reviews test failures and proposes rules). Flag rules that show signs of LLM confabulation rather than verified Cosmos DB knowledge:
- ๐ด Claims specific RU costs, internal limits, or SDK implementation details without a verifiable documentation link
- ๐ด Attributes a test failure to a Cosmos DB "best practice" that may actually be a framework bug, test misconfiguration, or code generation error
- ๐ด Invents SDK method signatures, parameters, or behaviors that don't exist in official SDK docs
- ๐ด Presents a scenario-specific workaround as a universal rule (e.g., "always do X" when X only applies to the exact test that failed)
- ๐ก Rule has no Reference link to official Microsoft documentation โ high hallucination risk
- ๐ก Rule's Incorrect/Correct examples look synthetic rather than drawn from real SDK usage patterns
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 ยท 75 lines ยท 860 tokens per session scan A cd6bfa3816f8
cosmosdb-agent-kit code-review.instructions.md is an instructions file published in the GitHub repository AzureCosmosDB/cosmosdb-agent-kit (53 stars, last pushed 10d ago), licensed MIT. It adds 860 tokens to every session, about $0.0043 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
postgres-skills AGENTS.md
Instructions for neondatabase/postgres-skills, covering agents.md, repository overview, creating a new skill, directory structure and naming conventions.
seekstone CLAUDE.md
Instructions for shaqmughal/seekstone, covering claude.md, what this repo is, commands, the harness itself (run after npm install) and architecture.
plan-forge database.instructions.md
Database patterns for .NET โ Dapper/EF Core, parameterized queries, migration strategy.
js-bson CLAUDE.md
Instructions for mongodb/js-bson, a project described as: BSON Parser for node and browser.
pg-dash CLAUDE.md
Claude Code instructions for indiekitai/pg-dash, covering pg-dash claude.md, ้กน็ฎ็ปๆ, ๆๅปบไธๆต่ฏ, ๅ็็บชๅพ๏ผๅผบๅถ๏ผ and ๅ็ๆฃๆฅๆธ ๅ๏ผๆฏๆฌก npm publish ๅๅฟ ้กปๆๅบๆง่ก๏ผ.
jdbc-driver CLAUDE.md
Claude Code instructions for keboola/jdbc-driver, covering claude.md, project overview, build & test commands, jdbc driver and vscode sqltools extension.