steelmorgan/agent-development-ext
framework_eng/skills/framework-meta/agent-development-ext/SKILL.md
1C BSL Framework extension for agent-development skill. Use together with the base agent-development skill when creating or modifying framework agents (analyst, architect, developer, reviewer, tester, explorer). Covers: universal agent format (Cursor + Claude Code), model tier mapping, framework-specific frontmatter fields, 1C BSL domain context.
$ install --global
skillsauthnpx skillsauth add steelmorgan/1c-agent-based-dev-framework agent-development-extInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
Security Scan Results
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
3
Scanners Passed
9
Scanners in report
Clean
TrivyContainer and dependency vulnerability scanner
95%
Clean
SemgrepStatic code analysis for vulnerabilities
95%
Clean
mcp-scan (Snyk)Model Context Protocol security validation
95%
Skipped
Snyk (dep)Open source security scanning
50%
Skipped
Socket.devSupply chain security analysis
50%
Skipped
VirusTotalMulti-engine malware detection
50%
Skipped
CrowdStrikeAdvanced threat intelligence
50%
Skipped
OSV-ScannerOpen Source Vulnerability database check
50%
Skipped
OWASP Dep-Check
50%
Last scanned: May 26, 2026, 5:07 AM16.3s1 file scanned
SKILL.md
- name:
- agent-development-ext
- description:
- >
- Covers:
- universal agent format (Cursor + Claude Code), model tier mapping,
Agent Development — 1C BSL Framework Extension
Base skill:
agent-development(Anthropic). Read the base skill first — it contains the general principles for creating agents. This file adds only 1C-specific details and adapts them for our framework.
1. Universal Framework Agent Format
A single .md file works for both Cursor and Claude Code without transformation.
Frontmatter
---
name: agent-name # lowercase, hyphens, 3-50 chars
description: >
One-liner + trigger conditions.
Use proactively when...
<<example>>
Context: ...
user: "..."
assistant: "..."
<<commentary>>...<</commentary>>
<</example>>
model: sonnet # haiku | sonnet | opus (алиас, CLI подставит конкретную модель для Cursor)
readonly: true # true для read-only агентов (analyst, explorer, reviewer)
skills: # Claude Code подгрузит автоматически; Cursor проигнорирует
- spec-standard
- search-before-write
---
Body (System Prompt)
Written in the second person (You are...). Structure:
You are [role] specializing in [domain] for 1C:Enterprise (BSL).
**Skills and Rules (for Cursor):**
- `skill-name` — short purpose
- `rule-name` — short purpose
**Your Core Responsibilities:**
1. [Responsibility 1]
2. [Responsibility 2]
**Input:**
- [What the agent receives]
**Output:**
- [What the agent produces]
**Protocol:**
1. [Step 1]
2. [Step 2]
**Quality Standards:**
- [Criterion 1]
- [Criterion 2]
**Boundaries:**
- [What the agent DOES NOT do]
The "Skills and Rules" section in the body is needed for Cursor — it ignores the skills field in frontmatter. We duplicate only the names and a brief purpose there.
2. Framework Roles and Models
Role → Model Mapping Table
| Role | model | readonly | Rationale | |------|-------|----------|-----------| | explorer | haiku | true | Deterministic work, tools deliver precise answers | | analyst | sonnet | true | Requirements analysis and specification writing | | tester | sonnet | false | Writing and executing tests | | architect | sonnet | true | Technical decisions and trade-offs | | developer | sonnet | false | Implementing code, following TDD | | reviewer | opus | true | Critical role — assessing artifacts, tier ≥ author |
Reviewer Rule
The reviewer model MUST be at least the author’s model. If the author is sonnet, the reviewer must be sonnet or opus.
CLI: Model Mapping
Defaults live in tools/model-defaults.json. When running python tools/install.py, the user can accept defaults, pick per-agent (mode [a]), or use non-Anthropic models.
3. Cursor / Claude Code Compatibility
| Field | Claude Code | Cursor |
|------|-------------|--------|
| name | ✓ | ✓ |
| description | ✓ trigger + examples | ✓ description as the rule text |
| model | ✓ alias | ✓ CLI writes the concrete model |
| readonly | — (tools/disallowedTools) | ✓ supported natively |
| skills | ✓ preloads | ✗ ignored → mirror names in the body |
| color / tools | ✓ | ✗ |
Unknown fields are simply ignored — not an error.
4. 1C BSL Domain: Context for the System Prompt
When writing the system prompt for 1C agents, keep these points in mind:
Key Constraints
- The agent DOES NOT create metadata objects — only code in
.bslmodules. - BSL spans server and client modules with compilation directives (
&НаСервере,&НаКлиенте). - Tools are discovered through MCP (
tools/list); the "capability" section is not needed in the agent file. - Skills under
tool-usage/*explain when and how to use specific MCP tools. - Standards: MADR 4.0, RFC 2119, YaxUnit, БСП.
5. Framework Agent Creation Checklist
- [ ]
name— lowercase, hyphens, 3-50 chars - [ ]
description— trigger conditions + 2-3<<example>>blocks - [ ]
model— alias from the table (haiku/sonnet/opus) - [ ]
readonly— true for read-only roles - [ ]
skills— list of skills to preload - [ ] Body — system prompt in second person (You are...)
- [ ] In the body — a "Skills and Rules" section with names and purposes
- [ ] In the body — Core Responsibilities, Protocol, Quality Standards, Boundaries
- [ ] No "capability" section — tools come through MCP
- [ ] No separate "Input/Output" tables — keep them inline
depends_on: []
Related Skills
steelmorgan/vanessa-diagnostics
tools
VerifiedTrustedCommunity
Diagnostics for Vanessa Automation runs. Use when a feature scenario failed, artifacts were not created, or you need to classify a failure after launch.
65SKILL.mdUpdated May 26, 2026
steelmorgan/vanessa-authoring
tools
VerifiedTrustedCommunity
Creating and refining Vanessa Automation feature scenarios based on real project requirements. Use when you need to write or update a scenario test, not just run it.
65SKILL.mdUpdated May 26, 2026
steelmorgan/framework_eng/skills/tool-usage/v8-session-manager
tools
VerifiedTrustedCommunity
--- name: v8-session-manager description: Use when working with the 1С session manager (v8-session-manager) - launch, configuration, connecting 1С clients, reading session_list, calling proxied MCP-tools from 1С extensions, diagnostics. Triggers: mention of `v8-session-manager`, `session_list`, 1С extension MCP showcase, error “no active sessions” / “session_id required”, connecting a client to the manager via `mcpMode=ws`. provides_capabilities: # Built-in manager tools — always available whi
65SKILL.mdUpdated May 26, 2026
steelmorgan/v8-runner
tools
VerifiedTrustedCommunity
Use when Codex needs to manage v8-runner on local 1C projects through the CLI: configure v8project.yaml, initialize infobases or EDT workspaces, build sources from Designer or EDT, run syntax checks and tests, dump infobase changes, convert source formats, load or export artifacts, launch 1C clients, or choose safe 1C automation command sequences.
65SKILL.mdUpdated May 26, 2026