# Nop.Plugin.Misc.AIPlugin — AI Agent Protocol This plugin follows the AI Agent Core Protocol defined in AyCode.Core. Same rules apply: `[LOADED_DOCS]` prefix (count+delta format), docs-before-code, no-re-read policy, context recovery on compaction, explicit consent for modifications. **Full protocol:** `../../../../../../Aycode/Source/AyCode.Core/.github/copilot-instructions.md` **Protocol history:** `../../../../../../Aycode/Source/AyCode.Core/.github/LLM_PROTOCOL_DECISIONS.md` ## Session Setup **Mandatory reads at session start** — in addition to this `copilot-instructions.md`, the agent MUST also load: 1. **Canonical AI Agent Core Protocol** — AyCode.Core's `copilot-instructions.md` (for the full numbered Rules #1-5 that this inherit file references but does NOT duplicate). Path: `../../../../../../Aycode/Source/AyCode.Core/.github/copilot-instructions.md`. 2. **The four workspace skills** (see `## Shared Agent Skills` section below for paths): - `docs-discovery/SKILL.md` — **reactive** (triggers on domain questions) - `docs-check/SKILL.md` — **reactive** (triggers at end of code-modifying responses) - `protocol-audit/SKILL.md` — **on-demand** (triggers on explicit "audit protocol" command) - `adr-author/SKILL.md` — **on-demand + LLM-suggested** (triggers on explicit planning/design requests, or LLM flag + user confirmation) The first response's `[LOADED_DOCS: ...]` prefix must reflect **6 files** (this file + AyCode.Core's `copilot-instructions.md` + 4 SKILL.md files). **Why mandatory**: workspace skills are NOT in Claude Code's native skill-registry / system-reminder — without pre-loading, implicit triggers (e.g., "domain question → invoke docs-discovery" at the moment of the question) may be missed. This file inherits rules from AyCode.Core, so the canonical protocol file must be loaded alongside. One-time cost per session; Rule #3 (no-re-read) prevents repeated reads; re-read only if Rule #4 (Context Recovery) fires. **Amortization — critical, do NOT re-evaluate per-turn**: the Session Setup cost is measured over the ENTIRE session, not per single turn. A typical session has many turns; the first domain question alone already recoups the investment (alternative — repeated source-code `Grep`/`Read` per turn — costs 10-20K tokens *per turn* with lower output quality). Do NOT flag pre-loaded content as "wasteful" for turns that don't invoke it — the design depends on cross-turn amortization + Rule #3 (no-re-read) + on-demand specificity of Rule #4 (Context Recovery). This amortization is the **central token-economics principle** of the entire protocol stack. ## Workspace Dependencies @repo { name = "Nop.Plugin.Misc.AIPlugin" type = "consumer" layer = 3 own-dep-repos = [ "Mango.Nop.Core: ../../Libraries/Mango.Nop.Core", "AyCode.Core: ../../../../../../Aycode/Source/AyCode.Core" ] } > nopCommerce plugin (Layer 3 consumer) built on Mango.Nop.Core (Layer 2 framework) and AyCode.Core (Layer 0). Contains FruitBank-specific AI/server-side logic for the nopCommerce plugin surface. > External repos in `own-dep-repos` are fully accessible — read their source code, docs, and `.github/copilot-instructions.md` freely when you need type definitions, base classes, or context. ## Shared Agent Skills Skills defined in other repos. **All four are pre-loaded at session start per the `## Session Setup` section above** (mandatory — ensures implicit triggers fire reliably): - **docs-discovery** — Load relevant `.md` documentation (main + `_ISSUES` + `_TODO` paired sets) BEFORE source-code search or modifications. Saves tokens vs. grep-based rediscovery. Location: `AyCode.Core/.github/skills/docs-discovery/SKILL.md` (see `own-dep-repos` above for the relative path to AyCode.Core) **Invoke proactively** before any domain-related coding task (see "Documentation-first coding" below). Honours the active repo's **no-re-read** rule. - **protocol-audit** — Cross-repo consistency audit for `.github/copilot-instructions.md` across all repos. Location: `AyCode.Core/.github/skills/protocol-audit/SKILL.md` Activate from an AyCode.Core session, or read the SKILL.md directly and follow its steps. - **docs-check** — Evaluate loaded `.md` files at the end of every code-modifying response: drift vs code, missing topic coverage, csproj-glob registration gaps for new `.md` files, and new issue/TODO candidates. Emits the `[DOCUMENTATION CHECK]` section. Location: `AyCode.Core/.github/skills/docs-check/SKILL.md` **Invoke at the end of every code-modifying response.** Read-only on loaded docs; all patches surface as proposals (Rule #5 approval required). - **adr-author** — Create Architecture Decision Records (ADRs) for architecturally significant design decisions. Structured interview (context → alternatives → trade-offs → decision → consequences) producing a durable `docs/adr/NNNN-.md` file (product decisions) or a new `LLMP-DEC-N` row in the protocol decision log (meta-protocol decisions). Location: `AyCode.Core/.github/skills/adr-author/SKILL.md` **Invoke on explicit user request** ("let's plan X", "decide Y vs Z", "design the W module") **or proactively flag** when the conversation looks ADR-worthy (user must confirm — never auto-invoke). ## Protocol History Cumulative log of LLM-protocol decisions (rule changes, new skills, structural shifts): - `AyCode.Core/.github/LLM_PROTOCOL_DECISIONS.md` Read this file when you need to understand **why** a rule is the way it is, or before proposing a protocol change — it may save a debate about something already resolved. ## Documentation-first coding Before running any source-code `Grep` / `get_file` / `code_search` in response to a domain-related request, invoke the **`docs-discovery`** skill (path above). Scans `docs/` folders in THIS repo AND in referenced repos (via `own-dep-repos`) via Glob, loads paired .md sets as a unit. Rule-number-agnostic — refers to rule NAMES (no-re-read, folder-navigation, explicit-consent) which are stable across repos. ## Related Documentation | Document | Topic | |---|---| | `docs/` | Plugin-specific documentation (if present) | | `README.md` | Plugin overview | | `plugin.json` | nopCommerce plugin manifest |