v8std standards harness
v8std-standards-harness
A skill and validator that stop an agent from silently skipping the v8std 1C development standards check on BSL code
Low risk
We rate an entry low when it mostly gives the agent instructions and reference material.
Why this level
- Only reads the standards reference and leaves text evidence lines, does not change configuration code
- In gate mode it can block a commit with zero evidence lines, which is intentional default behavior
Install
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/v8std-standards-harnessDetects the agents on your machine, checks the risk and pins the version.
Other ways to install
Run in a terminal in the project folder
npx skills add axelboman277/v8std-standards-harness --skill v8std-standards-harness -a claude-code -yThe skills tool installs the current version from the repository. Add the -g flag to use the skill in every project.
Run in a terminal in the project folder
npx skills add axelboman277/v8std-standards-harness --skill v8std-standards-harness -a cursor -yThe skills tool installs the current version from the repository. Add the -g flag to use the skill in every project.
Run in a terminal in the project folder
npx skills add axelboman277/v8std-standards-harness --skill v8std-standards-harness -a github-copilot -yThe skills tool installs the current version from the repository. Add the -g flag to use the skill in every project.
Run in a terminal in the project folder
npx skills add axelboman277/v8std-standards-harness --skill v8std-standards-harness -a codex -yThe skills tool installs the current version from the repository. Add the -g flag to use the skill in every project.
Run in a terminal in the project folder
npx skills add axelboman277/v8std-standards-harness --skill v8std-standards-harness -a gemini-cli -yThe skills tool installs the current version from the repository. Add the -g flag to use the skill in every project.
Run in a terminal in the project folder
npx skills add axelboman277/v8std-standards-harness --skill v8std-standards-harness -a cline -yThe skills tool installs the current version from the repository. Add the -g flag to use the skill in every project.
Run in a terminal in the project folder
npx skills add axelboman277/v8std-standards-harness --skill v8std-standards-harness -a roo -yThe skills tool installs the current version from the repository. Add the -g flag to use the skill in every project.
A fork of Roo Code, same .roo folders.
Run in a terminal in the project folder
npx skills add axelboman277/v8std-standards-harness --skill v8std-standards-harness -a opencode -yThe skills tool installs the current version from the repository. Add the -g flag to use the skill in every project.
Connect the standards MCP service by adding { "mcpServers": { "v8std": { "type": "http", "url": "https://ai.v8std.ru/mcp" } } } to your config, copy SKILL.md and the references folder into your skills directory, check the situation map while working on BSL code and leave machine-readable evidence lines, then run node tools/v8std-evidence-validator.mjs validate-pack . --gate before committing.
Other ways from the author
{ "mcpServers": { "v8std": { "type": "http", "url": "https://ai.v8std.ru/mcp" } } }Added to .mcp.json; for Codex the same URL goes into ~/.codex/config.toml.
This is third-party code. Review the repository files before installing.
What it does
The skill connects to the external v8std standards MCP service and solves one problem: making a skipped check visible instead of silent. A two-way situation map, by metadata object and by code action, points the agent to standard numbers for common cases. When a situation is not on the map, the skill requires at least one v8std_search call per work phase. A sentinel check confirms the standards service is alive and its pointer is current, falling back to an offline cache with an honest skipped mark and reason when it is not. A separate Node validator checks that machine-readable evidence lines were actually left, and can run in lint mode for a one-off pass or gate mode to block a commit that has zero entries.
Who it is for. For 1C developers who let an AI agent touch production BSL code and want verifiable proof that development standards were actually applied, not just mentioned in an instruction.
Good fit when
- An agent writes or reviews queries, metadata objects, document posting or 1C configuration extensions
- You need machine-checkable proof of a standards check before a commit
- You need a clear map of which standards apply to a given situation instead of hoping the model recalls all three hundred of them
Not a fit when
- The project is not on 1C: the standards maps are specific to the platform
- There is no access to https://ai.v8std.ru/mcp and no local cache was built per the instructions
- You need a static analyzer or syntax checker: the harness only checks that an evidence trail exists, not the quality of the decision
Example request
Before committing, run the v8std standards check in gate mode against the changed configuration objectsLimitations
The validator runs offline and checks only the format of evidence lines, not their truthfulness: a made-up standard id will formally pass. The sentinel check catches an unreachable service or its pointer rolling back, but not the map itself falling behind the service. The map is a snapshot dated by last_reviewed_at, and keeping it current is the repository owner's responsibility. The repository does not contain the text of the 1C standards themselves, only ids and titles; the actual requirements are read via v8std.ru or ITS. Node.js 18 or newer is needed for the validator.
How to disable. Remove the skill from the agent's skills directory and delete the v8std server from your MCP client configuration.
Security check
- Only reads the standards reference and leaves text evidence lines, does not change configuration code
- In gate mode it can block a commit with zero evidence lines, which is intentional default behavior
README in short
The README describes the problem of a completed standards check being indistinguishable from a skipped one, and a four-layer solution: a situation map, mandatory search, a service sentinel check and machine validation of evidence lines. It includes a layer table, an example of four evidence-line types (applied, skipped, discovered, sentinel), a quick start covering the MCP connection and validator commands, the difference between lint and gate modes, and a separate honesty section: the harness does not analyze code, does not verify that standards exist, and does not guarantee the map is current. It credits the standards text to its 1C rights holder, noting the repository only holds ids. MIT license on the repository's own code.
SKILL.md
--- name: v8std-standards-harness description: > Use when writing, reviewing, or designing 1C:Enterprise (BSL) code and configuration metadata — enforces a fail-closed check against official 1C development standards (v8std) instead of relying on the agent to remember them. Triggers on: writing a query, creating or changing a metadata object (catalog, document, register, form, role), document posting and register movements, export procedures, client-server transitions, privileged mode, localization, configuration extensions, and any BSL code review. Produces machine-readable evidence lines that a validator can gate on, so a skipped standards check cannot pass silently. Not for non-1C codebases. --- # Обвязка для работы со стандартами 1С (v8std) Обвязка вокруг MCP-сервиса стандартов разработки 1С (https://ai.v8std.ru/mcp). Решает одну конкретную задачу: не дать агенту молча пропустить сверку со стандартами.
FAQ
What if a standard is not on the situation map?
The skill requires at least one v8std_search call per work phase, so the map does not freeze at the moment it was created.
What is the difference between lint and gate mode?
lint only checks the formatting of found evidence lines and treats zero entries as a clean pass. gate requires at least one entry of any type and turns zero entries into a failure, which makes it suitable for blocking a commit.
Related
A skills library that gives coding agents a development process: brainstorming, planning, TDD, subagents and code review
Skills for real engineers by Matt Pocock
Skills For Real Engineers
Small composable skills for engineering with agents: plan grilling, TDD, bug diagnosis, code review and architecture
GitHub toolkit for spec-driven development: the specify CLI adds agent commands and skills to a project, from principles to implementation
Reference MCP servers
Model Context Protocol servers
Official reference MCP servers: Filesystem, Fetch, Git, Memory, Sequential Thinking, Time and Everything