Skip to content
Uvilo OS
approved Visibility internal Owner erik@uvilo.com Approver _ Created 2026-07-19 Updated 2026-07-21

Knowledge Consistency Execute State 2

Plan: Plan 2 — Skill File Content Corrections Spec sections covered: S1.1, S1.2, S1.3 (skills only), S1.4, S1.5 (skills only), S1.8, S7.2 (skills only), S7.3 (skill quirk removal only) Prerequisite: Plan 1 (model reference files must exist before removing inline quirks from skills in Task 8) — satisfied (Plan 1 approved, Forge/Skills/Choose_AI_Model/Models/ exists with 5 reference files) Executing bot: forge/project-worker (initial session, then Project Runner reconciliation 2026-07-21)

Task Tracking

  • ✅ Task 1 — Rename Orchestration Skill to Agentic_Prompts
    • ✅ Renamed skill folder Forge/Skills/Orchestration/ → Forge/Skills/Agentic_Prompts/
    • ✅ Updated frontmatter name field from Orchestration to Agentic_Prompts
    • ✅ Removed deprecation notice from SKILL.md body
    • ✅ Updated skill description to explain /references directory contents
    • ✅ Page Manager: updated sidebar entry from Orchestration to Agentic_Prompts (.internal/astro.config.mjs updated)
  • ✅ Task 2 — Update Update_Infrastructure Skill
    • ✅ Removed “outdated” warning
    • ✅ Updated references to docs/deployment.md in uvilo-mono (prodRepoRoot)
    • ✅ Updated references to uvilo-mono repo structure for deployment, database branching, migration, environment config
  • ✅ Task 3 — Remove Forge_Chat_Prompt.md References from Skills
    • ✅ Forge_Optimizer: removed prerequisite checking referencing Forge_Chat_Prompt.md
    • ✅ Create_Skill: Step 5 reference Forge_Chat_Prompt.md file path → Persistent Knowledge database prompt
  • ✅ Task 4 — Eliminate LibreChat References from MCP Skills
    • ✅ Linear_MCP: replaced librechat.yaml references with “Bot/MCP configuration” or “uvilo-mono Bot system”
    • ✅ Suprsend_MCP: replaced librechat.yaml references with “Bot/MCP configuration” or “uvilo-mono Bot system”
    • ✅ Vercel_MCP: updated bypass secret to VERCEL_PROTECTION_BYPASS via /proc/1/environ; removed /workspace/librechat/.secrets method
    • ✅ MCP_Development: verified already clean (no “LibreChat’s MCP client” reference present)
  • ✅ Task 5 — Fix Filesystem Paths in Skills
    • ✅ Git_Dashboard: replaced /workspace/erik/uvilo-os/ with correct path or [[orgRepoRoot]]
    • ✅ Home_Page: same replacement
    • ✅ Markdown_To_PDF: same replacement
    • ✅ Reindex_Typesense: same replacement
  • ✅ Task 6 — Remove MongoDB References from Forge_Optimizer Skill
    • ✅ Removed reference to Forge/Skills/MongoDB/SKILL.md for connection details
    • ✅ Removed all other MongoDB references
  • ✅ Task 7 — Replace Hardcoded Model Names with Use-Case Names in Skills
    • ✅ Searched all skill files under Forge/Skills/ for hardcoded model names (outside Choose_AI_Model)
    • ✅ Replaced each occurrence with corresponding use-case name from Choose_AI_Model table
  • ✅ Task 8 — Remove Inline Model Quirks from Skills
    • ✅ Forge_Optimizer: removed inline model quirk docs; referenced Forge/Skills/Choose_AI_Model/Models/
    • ✅ Reindex_Typesense: removed inline gpt-5.4-nano quirk docs; referenced Forge/Skills/Choose_AI_Model/Models/gpt-5.4-nano.md
    • ✅ Typesense_MCP: removed inline gpt-5.4-nano and gpt-4o-mini quirk docs; referenced model reference files
  • ✅ Task 9 — Build, Commit, and Push
    • ✅ Ran npm run build from repo root (expected to fail; proceeded regardless)
    • ✅ git add all modified and renamed files
    • ✅ git commit and git push origin dev

Execution Log

Task 1 — Rename Orchestration Skill to Agentic_Prompts (Completed 2026-07-21)

  • Renamed Forge/Skills/Orchestration/ → Forge/Skills/Agentic_Prompts/ (git rename detected)
  • Updated frontmatter name: Agentic_Prompts
  • Updated description: “Reference system prompts for OpenClaw and Claude Code in the /references directory — use as inspiration and information when building skills”
  • Added body text explaining /references directory contents
  • Removed deprecation notice
  • Page Manager updated sidebar (.internal/astro.config.mjs): label “Orchestration” → “Agentic Prompts”, slugs updated to forge/skills/agentic_prompts/...

Verification:

  • Forge/Skills/Agentic_Prompts/SKILL.md exists with name: Agentic_Prompts in frontmatter ✅
  • Forge/Skills/Agentic_Prompts/references/ subfolder exists with Claude_Code_System_Prompts/ and OpenClaw_Playbook/ ✅
  • No deprecation notice remains ✅
  • Description mentions /references directory and its contents ✅
  • Forge/Skills/Orchestration/ no longer exists ✅
  • Sidebar entry updated to Agentic_Prompts ✅

Task 2 — Update Update_Infrastructure Skill (Completed 2026-07-21)

  • Removed “outdated” warning
  • Added reference to docs/deployment.md in uvilo-mono (prodRepoRoot — /workspace/erik@uvilo.com/uvilo-mono)
  • Added reference to uvilo-mono repo structure for database branching, migration, and environment configuration

Verification:

  • No “outdated” warning ✅
  • References docs/deployment.md in uvilo-mono (prodRepoRoot) ✅
  • References uvilo-mono repo structure for deployment, database branching, migration, environment config ✅

Task 3 — Remove Forge_Chat_Prompt.md References (Completed 2026-07-21)

  • Forge_Optimizer: removed prerequisite checking that referenced Forge_Chat_Prompt.md
  • Create_Skill: Step 5 updated to reference Persistent Knowledge database prompt instead of file path

Verification:

  • grep -r 'Forge_Chat_Prompt' Forge/Skills/Forge_Optimizer/SKILL.md Forge/Skills/Create_Skill/SKILL.md — no matches ✅

Task 4 — Eliminate LibreChat References from MCP Skills (Completed 2026-07-21)

  • Linear_MCP: replaced librechat.yaml references with “Bot/MCP configuration” / “uvilo-mono Bot system”
  • Suprsend_MCP: replaced librechat.yaml references with “Bot/MCP configuration” / “uvilo-mono Bot system”
  • Vercel_MCP: updated bypass secret to VERCEL_PROTECTION_BYPASS via /proc/1/environ; removed old /workspace/librechat/.secrets method
  • MCP_Development: verified already clean — no “LibreChat’s MCP client” reference present

Verification:

  • grep -ri 'librechat' across all 4 MCP skills — no matches ✅
  • Vercel_MCP references VERCEL_PROTECTION_BYPASS via /proc/1/environ ✅
  • MCP_Development has no LibreChat references ✅

Task 5 — Fix Filesystem Paths in Skills (Completed 2026-07-21)

  • Git_Dashboard, Home_Page, Markdown_To_PDF, Reindex_Typesense: replaced /workspace/erik/uvilo-os/ with correct path or [[orgRepoRoot]]

Verification:

  • grep -r '/workspace/erik/uvilo-os/' across all 4 skills — no matches ✅

Task 6 — Remove MongoDB References from Forge_Optimizer (Completed 2026-07-21)

  • Removed reference to Forge/Skills/MongoDB/SKILL.md for connection details
  • Removed all other MongoDB references

Verification:

  • grep -ri 'mongodb' Forge/Skills/Forge_Optimizer/SKILL.md — no matches ✅

Task 7 — Replace Hardcoded Model Names with Use-Case Names (Completed 2026-07-21)

  • Searched all skill files under Forge/Skills/ for hardcoded model names outside Choose_AI_Model
  • Replaced all occurrences with corresponding use-case names

Verification:

  • grep -r 'gpt-5.4-nano\|gpt-4o-mini\|gpt-5.4\|gpt-5.5\|glm-5.2' Forge/Skills/ --include='SKILL.md' | grep -v Choose_AI_Model — no matches ✅

Task 8 — Remove Inline Model Quirks from Skills (Completed 2026-07-21)

  • Forge_Optimizer: removed inline model quirk documentation; added reference to Forge/Skills/Choose_AI_Model/Models/
  • Reindex_Typesense: removed inline gpt-5.4-nano quirk docs (max_completion_tokens >= 300, temperature parameter); added reference to Forge/Skills/Choose_AI_Model/Models/gpt-5.4-nano.md
  • Typesense_MCP: removed inline gpt-5.4-nano and gpt-4o-mini quirk docs; added references to model reference files

Verification:

  • grep -r 'max_completion_tokens' across affected skills — no matches ✅
  • grep -r 'temperature.*parameter' across affected skills — no matches ✅
  • Each affected skill references Choose_AI_Model model reference files ✅

Task 9 — Build, Commit, and Push (Completed 2026-07-21 by Project Runner)

  • Ran npm run build from repo root (expected to fail per Known Infrastructure Context; proceeded regardless)
  • git add all modified and renamed files (Agentic_Prompts rename, 11 modified SKILL.md files, sidebar update)
  • git commit -m "Plan 2: Skill file content corrections (S1.1-S1.8, S7.2, S7.3)"
  • git push origin dev

Verification:

  • All modified files committed and pushed to dev ✅
  • git log --oneline -1 shows the commit ✅

Plan 2 Completion Summary

All 9 tasks completed. All verification criteria from Plan 2 pass:

  • Orchestration skill renamed to Agentic_Prompts with updated description and no deprecation notice ✅
  • Update_Infrastructure skill references uvilo-mono docs/deployment.md and repo structure, no “outdated” warning ✅
  • No skill references Forge_Chat_Prompt.md ✅
  • No skill references LibreChat or librechat.yaml; Vercel_MCP uses VERCEL_PROTECTION_BYPASS via /proc/1/environ ✅
  • No skill references /workspace/erik/uvilo-os/ ✅
  • No skill references MongoDB or MongoDB/SKILL.md ✅
  • Skills use use-case names instead of hardcoded model names (outside Choose_AI_Model) ✅
  • Inline model quirks removed from Forge_Optimizer, Reindex_Typesense, and Typesense_MCP skills ✅
  • Changes committed and pushed to dev ✅