Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 0 additions & 1 deletion README.ko.md
Original file line number Diff line number Diff line change
Expand Up @@ -637,7 +637,6 @@ name: agent-name
description: 에이전트의 역할
model: sonnet
tools: Read, Edit
temperature: 0.3
---
```

Expand Down
1 change: 0 additions & 1 deletion README.md
Original file line number Diff line number Diff line change
Expand Up @@ -663,7 +663,6 @@ name: agent-name
description: What the agent does
model: sonnet
tools: Read, Edit
temperature: 0.3
---
```

Expand Down
4 changes: 4 additions & 0 deletions docs/CUSTOM_EXTENSIONS.md
Original file line number Diff line number Diff line change
Expand Up @@ -214,6 +214,10 @@ The eight agent definitions exist in two layers — `project/.claude/agents/` an

**CI enforcement**: `.github/workflows/validate-skills.yml` runs `scripts/check_agents.sh` (PowerShell twin: `scripts/check_agents.ps1`). It strips frontmatter and normalizes that single sentence, then fails (exit 2) if the agent **bodies** otherwise diverge — so a behavioral instruction edited in one layer but not the other cannot ship silently.

**Advisory frontmatter (`applies_to`, `keywords`)**: every agent declares `applies_to` (a glob list) and `keywords`. These are **not** official Claude Code runtime path-triggers — the runtime delegates to a sub-agent only by its `description`. They are advisory inputs consumed by the `fleet-orchestrator` skill's Top-K routing score (`score = 2 * matched_applies_to_globs + 1 * matched_keywords`, see `global/skills/_internal/fleet-orchestrator/SKILL.md`). Opening a file that matches an `applies_to` glob does **not** auto-invoke the agent; keep delegation intent in `description`.

**`structure-explorer` memory scope**: `structure-explorer` is intentionally the only agent with `memory: local` and no `initialPrompt` — it performs stateless, 1-shot, breadth-first exploration with no cross-call state to persist, so the project-memory bootstrap pattern used by the other seven agents does not apply.

**Skill reference files are intentionally NOT guarded this way.** Beyond the four workflow references above, the `plugin/skills/*/reference/` tree is a curated *re-structuring* of `project/.claude/rules/` — content is split and recombined across files (e.g. `rules/api/observability.md` becomes `observability.md` + `logging.md`; `rules/coding/` standards fan out into `general.md`/`quality.md`/`concurrency.md`/`memory.md`). A per-file 1:1 drift check would false-positive on that reorganization, so the plugin reference bundle is maintained as an independent curated distribution.

### Version Declarations (VERSION_MAP SSOT)
Expand Down
Loading