Windsurf rules

1---
2name: delegate-task
3model: sonnet
4description: "Delegate one complex task to a single subagent, review its work in two stages before merging back. Sequential — one agent at a time, with oversight. Use when a task is complex and requires careful review before the result is accepted. Distinct from dispatch-agents (no parallelism here; reviewer sees full diff before proceeding)."
5---
6 
7# story: e45s30
8 
9# Delegate Task
10> **HARD GATE** — **HARD GATE** — Delegated work must have clear success criteria and verification commands. The delegate must be able to verify completion independently.
11 
12 
13Delegate a single complex task to a subagent with a two-stage review gate before accepting the result. Use when oversight of a single task matters more than speed.
14 
15**Distinct from `dispatch-agents`:** This skill runs one subagent sequentially with a mandatory review. `dispatch-agents` runs multiple subagents in parallel without inter-task review gates.
16 
17## Subagent depth tiers (e45s30)
18 
19Select brief depth from task `risk:` and skill `effort:` before spawning:
20 
21| Tier | When | Brief includes |
22|------|------|----------------|
23| `full_maturity` | P0 stories, multi-file refactors, security work | Full template + CONVENTIONS excerpts + threat model if present |
24| `standard` | Default implementation tasks | Goal, scope, out-of-bounds, constraints, verify, prior decisions |
25| `minimal_decisive` | Light probes, read-only audits | Goal, verify, explicit file list (≤15 lines total) |
26 
27State `depth: <tier>` in the Agent tool description field.
28 
29## Process
30 
31### 1. Define the task
32 
33Before spawning the agent, read `specs/state.yaml` if it exists. Then write a minimal self-contained brief using this template (brief size directly controls token cost and hallucination risk — do not pad):
34 
35```
36Goal: [one sentence — specific, measurable outcome]
37In scope: [explicit file or module list]
38Out of bounds: [what NOT to do]
39Constraints: [relevant CONVENTIONS.md rules, existing patterns, test requirements]
40Verify: [runnable command]
41Prior decisions: [relevant entries from specs/state.yaml — omit section if none apply]
42```
43 
44Do not include full file contents, full conversation history, or decisions unrelated to this task.
45 
46### 2. Spawn the subagent (iterative retrieval, max 3 cycles)
47 
48Use the Agent tool with a **fresh context** per spawn. Pass prior decisions only via `specs/state.yaml`.
49 
50**Cycle:** dispatch → evaluate output vs goal → refine brief → re-spawn if needed (max 3 cycles).
51 
52Include in each brief:
53- All context the agent needs (it starts cold — no shared state)
54- Reference to CONVENTIONS.md constraints
55- The verify command it must run before reporting done
56 
57### 3. Stage 1 review — output inspection
58 
59When the subagent returns, review its report before looking at the diff:
60- Did it run the verify command? Did it pass?
61- Does it explain what it changed and why?
62- Are there any concerns raised by the agent?
63 
64If the report raises red flags, ask the subagent for clarification or re-run with adjusted instructions.
65 
66### 4. Stage 2 review — diff inspection
67 
68Inspect the actual diff:
69```bash
70git diff main...HEAD
71```
72 
73Check:
74- [ ] Changes are scoped to what was asked — nothing extra
75- [ ] No `any`, no `@ts-ignore`, no disabled lint rules
76- [ ] Tests added for new behavior
77- [ ] CONVENTIONS.md compliance (naming, structure, no gh issue creation)
78- [ ] Boy Scout Rule: touched areas are cleaner than before
79 
80### 5. Decision
81 
82- **Accept**: merge the result into the main working branch
83- **Revise**: send back to the subagent with specific feedback
84- **Reject**: discard and re-approach differently
85 
86**After accepting**, append to `specs/state.yaml` under `## Active Decisions`:
87```
88**[task short name]**: [what approach the agent chose and why — one sentence]
89```
90 
91Report the decision and rationale to the user.
92