Pack comparison
Two packs, side-by-side. Merged comparisons, shared shape, and diff highlights in one view.
advisor-pattern-v2
Sonnet executes, Opus advises. Route intelligence by confidence, pay Opus only when it matters.
Install
npx attrition-sh pack install advisor-pattern-v2Token budget
12,000
Pass rate
—
Avg tokens
—
Publisher
Agent Workspace
Compatibility
Contract summary
4 required outputs, 5 permissions, 5 completion conditions.
out: .advisor/<task-id>.json
subagent-delegation-three-isolation-modes
SkillTool vs AgentTool, 6 built-in types, 3 isolation modes. The knob every harness engineer gets wrong first.
Install
npx attrition-sh pack install subagent-delegation-three-isolation-modesToken budget
50,000
Pass rate
—
Avg tokens
—
Publisher
Agent Workspace
Compatibility
Contract summary
3 required outputs, 2 permissions, 3 completion conditions.
out: .sidechains/<task-id>.jsonl
Shared shape
What both packs have in common
Overlap across canonical pattern, compatibility, tags, and required packs.
Compatibility
Tags
Merged comparisons
Head-to-head claims from both packs
Each row is attributed to the pack that authored it. The winner column is normalised to this compare view (A / B / Tie).
| Source | Alternative | Axis | Winner | Note |
|---|---|---|---|---|
| A | single-model-sonnet | accuracy | A | Advisor retains ~93% of Opus-only pass-rate vs Sonnet-only's ~85%. The delta comes from the 8-12% of steps that actually need Opus. |
| A | single-model-sonnet | cost | Alternative | Sonnet-only is ~20% cheaper than the advisor variant on the same workload — if the quality gap doesn't matter, skip the advisor. |
| A | prompt-chaining | complexity | Alternative | Prompt chaining is simpler to implement but can't route by difficulty. Advisor adds complexity to capture the cost/quality tradeoff. |
| A | orchestrator-workers | latency | A | Advisor is a flat loop — no orchestrator fan-out. Lower tail latency on sequential tasks. Orchestrator wins on parallelisable ones. |
| B | advisor-pattern-v2 | cost | Tie | Advisor routes intelligence by confidence (cheap executor + expensive advisor). Subagent delegation routes context (keep wide reads out of the parent window). Different cost axes; stack them when both apply. |
| B | claude-code-guide | complexity | Alternative | Claude Code Guide is a 10-section reference covering the whole harness including subagents. This pack is the deep dive on just subagent delegation — use the guide for onboarding, this pack for production tuning. |
| B | turn-execution-pipeline | latency | Tie | Each subagent runs its own 9-step turn pipeline internally. Subagent spawns add 1-3s round-trip on top of a pipeline turn; fan-out parallelises across children to amortise. |
Diff highlights
What each pack brings that the other doesn't
Unique coverage and any measurable gap between the two.
Unique to A — Claude Advisor Pattern v2
Comparisons not in B
Compatibility A-only
Tags A-only
Unique to B — Subagent Delegation — Three Isolation Modes
Comparisons not in A
Compatibility B-only
Tags B-only
Token budget gap
Difference: 38,000 tokens. A (Claude Advisor Pattern v2) is the cheaper declared contract.
Swap / reset