steelmorgan/rlm-workflow
framework_eng/rules/rlm-workflow/SKILL.md
Before non-trivial domain work, read RLM
$ install --global
skillsauthnpx skillsauth add steelmorgan/1c-agent-based-dev-framework rlm-workflowInstall 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: Jun 30, 2026, 5:42 AM224.3s1 file scanned
SKILL.md
- name:
- rlm-workflow
- description:
- Before non-trivial domain work, read RLM
- alwaysApply:
- true
Memory Layout and Working with RLM
Two memory layers with different loading models. Native is push (always in context), RLM is pull (retrieved on demand). Do not confuse their purpose.
Layout (GUARD)
| Knowledge | Where | Why |
|---|---|---|
| Small, needed every turn, critical: safety, invariants, standing preferences, pointer to the active task | native (MEMORY.md + memory/*.md) | push - stays in context always, survives compaction for free |
| Universal, reusable, growing, detailed: patterns/antipatterns, architecture decisions with alternatives, stable domain facts about 1С/БСП, findings by modules | RLM (pull) | must NOT stay in context; retrieved by topic via semantic search |
| Transient task state: PENDING, next step, WIP | task_dir/.context/*.md (agent-context) | lives in the task, not in memory |
GUARD: do NOT keep universal knowledge in native/always-on context all the time (bloats every startup). Transient state is NOT in RLM.
TRIGGER - WRITE to RLM
- Learned a reusable pattern/antipattern (after >=2 iterations, see
skill-learning) - Made an architectural decision with alternatives
- Found a stable domain fact (platform/БСП behavior) that matters beyond the current task
→ apply the rlm-workflow skill (how to write: tool, level, causal decision).
TRIGGER - READ from RLM (at the boundary of work, NOT every turn)
- You are starting a non-trivial task or architectural decision in domain D
- You hit a problem that smells repetitive (bug, integration quirk)
→ one rlm_enterprise_context(query=<domain/symptom>) BEFORE design/implementation. Details are in the skill.
Parallel:
search-before-writechecks existing code; this trigger checks existing knowledge. The trigger is phase-based: at work entry, otherwise pull degrades into push.
MUST (invariant, always)
- Any write to RLM is only after
rlm_start_session(otherwise silent failure). - Universal knowledge is written to RLM, NOT duplicated in the native always-on core.
depends_on:
- rlm-workflow
- framework/rules/agent-context-protocol/SKILL.md
- framework/rules/search-before-write/SKILL.md
Related Skills
steelmorgan/onec-server-maintenance-hooks
development
VerifiedTrustedCommunity
1C server maintenance webhooks: container restart and external component cache cleanup
83SKILL.mdUpdated Jul 4, 2026
steelmorgan/dap-bsl-code-debug-procedure
development
VerifiedTrustedCommunity
Interactive DAP debugging of a single BSL procedure
83SKILL.mdUpdated Jul 4, 2026
steelmorgan/rlm-bsl-search
tools
VerifiedTrustedCommunity
Rules for using RLM tools for project search and navigation in 1C/BSL
83SKILL.mdUpdated Jul 4, 2026
steelmorgan/winow
development
VerifiedTrustedCommunity
Creates web applications and routes on Winow (a web server on OneScript and Autumn). Use when working with a web server on OneScript, routing, or Winow controllers.
83SKILL.mdUpdated Jul 4, 2026