메뉴
T-shirt sizing for tasks before dispatch. Runs heuristics and optionally LLM classification to determine task size (S/M/L/XL), recommend local vs cloud routing, and set tool budgets.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Branch, commit, push, and open PRs with humanized messages, then monitor GitHub Actions CI. Per-project autonomy gating; blocks direct pushes to the default branch and pushes with a dirty tree. Never auto-merges.
Dispatch patterns for Claude Code as the coding engine. Default harness. Implementation dispatches go through dispatch_coder.py (writes the dispatch receipt the commit gate requires).
Testing strategy, TDD enforcement, spec compliance, and regression checks.
Write implementation plans: bite-sized tasks for coding engine dispatch.
Automated fix loop for failed Quality/Reviewer checks. Parses failures, builds escalating prompts, dispatches through the active harness up to 3 times.
Run independent, file-disjoint plan tasks concurrently, each isolated in its own git worktree + branch. Collects per-task results; never auto-merges.
| name | complexity-triage |
| description | T-shirt sizing for tasks before dispatch. Runs heuristics and optionally LLM classification to determine task size (S/M/L/XL), recommend local vs cloud routing, and set tool budgets. |
| version | 1.0.0 |
| platforms | ["linux","macos","windows"] |
| metadata | {"hermes":{"tags":["triage","complexity","routing","dispatch","planning","t-shirt-sizing"],"related_skills":["local-model-router","writing-plans","architect","skill-discovery"]}} |
Run before planning to classify task complexity and set dispatch parameters.
Before every coding engine dispatch. The triage result determines:
terminal(
command="python3 ~/.hermes-coder/scripts/dynamic_curator.py --task '<task summary>' --repo '<project-dir>'",
workdir="~/.hermes-coder",
timeout=30
)
The script outputs JSON:
size: S, M, L, or XLconfidence: high, medium, or lowmethod: heuristic, llm, or fallbackrouting.engine: "local" or "cloud"routing.model: model identifiertool_budget: max_skills, max_mcp_tools, max_turnsskill_matches: top matched skills to inject| Size | Characteristics | Action |
|---|---|---|
| S | Single file, trivial verb, CSS-only, typo | Skip planning, raw dispatch, local model OK |
| M | Multi-file logic, new component, helper utility | Plan, selective skill injection, cloud model |
| L | New feature, major refactor, multi-file integration | Full plan, full skill injection, cloud model |
| XL | Architecture change, system redesign | Full plan, architect review, cloud model |
Every size dispatches to the coding engine. Size never licenses the coordinator to edit repo files itself — an S task is a small dispatch, not an inline edit. (2026-07-10: inline implementation by the coordinator burned two full turn budgets and shipped below the quality bar.)
The triage LLM pass runs through the active coding harness (resolved from --engine/coding.default_engine). If the harness is unavailable, heuristic-only triage still works (exit code 3). If heuristics are inconclusive and no LLM is available, the script defaults to M sizing with cloud routing.