subagent-driven-development

Use when executing implementation plans with independent tasks in the current session

Quality

65%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

High

Do not use without reviewing

Fix and improve this skill with Tessl

tessl review fix ./eval/local/skills/benchmarks/dependency/superpowers/subagent-driven-development/SKILL.md

The canonical home for this skill is subagent-driven-development in obra/superpowers

Quality

Content

81%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is a highly actionable, well-sequenced orchestration playbook with concrete commands, real bundled scripts, explicit validation gates, and feedback loops. Its main weakness is length and cross-section redundancy, plus two referenced prompt templates that are not present in the reviewed bundle.

Suggestions

Consolidate the repeated HEAD~1 / BASE warnings and the 'do not pre-judge findings' rule into a single authoritative location and reference it, rather than restating them in Handling Status, Constructing Reviewer Prompts, and Red Flags.

Ship implementer-prompt.md and task-reviewer-prompt.md alongside SKILL.md (or move them under references/) so the links in Prompt Templates resolve within the bundle.

Consider moving the Model Selection and Constructing Reviewer Prompts detail into a reference file, keeping SKILL.md as a tighter overview with the flowcharts and status table.

Dimension	Reasoning	Score
Conciseness	The body is dense with genuinely useful operational guidance rather than concepts Claude already knows, but at ~415 lines it repeats several points (the HEAD~1 warning, the 'do not pre-judge findings' rule, the no-parallel-implementers rule) across Handling Status, Constructing Reviewer Prompts, Red Flags, and Advantages, so it could be tightened.	3 / 5
Actionability	Concrete, executable guidance throughout: real script invocations ('scripts/review-package BASE HEAD', 'scripts/task-brief PLAN_FILE N'), exact git commands ('git merge-base main HEAD', 'git diff -U10'), a 5-element dispatch-prompt structure, and file-naming conventions that are copy-paste ready.	5 / 5
Workflow Clarity	Two dot flowcharts sequence the full process, the per-task loop has explicit validation gates (review clean before marking complete, re-review after fixes, final whole-branch review), feedback loops for error recovery, and a Red Flags checklist — and the batch/destructive cap does not apply because validation is pervasive.	5 / 5
Progressive Disclosure	Good structure with a dedicated Prompt Templates section and clearly signaled one-level-deep paths to scripts and prompt templates; the three referenced scripts are present in the bundle, but implementer-prompt.md and task-reviewer-prompt.md are referenced yet not verifiable in the bundle, a minor organization gap.	4 / 5
	Total	17 / 20 Passed

Description

48%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description leads with a strong explicit trigger but compresses the 'what' into the 'when' clause and never names the skill's distinguishing mechanism (subagent dispatch and review gates). It is distinguishable and triggerable, yet thin on stated capabilities.

Suggestions

Add a concrete 'what' clause naming the mechanism, e.g. 'Dispatches a fresh implementer subagent per task with a spec+quality review after each, and a broad final review. Use when executing implementation plans with independent tasks in the current session.'

Include a few natural user synonyms in the trigger, such as 'run this plan' or 'execute these tasks', alongside 'executing implementation plans'.

Name the per-task review gate in the description so the skill's value proposition is explicit rather than implied.

Dimension	Reasoning	Score
Specificity	The phrase 'executing implementation plans with independent tasks' names the domain but gives only a minimal, generic action ('executing') and omits the skill's concrete mechanism (dispatching subagents, per-task review).	2 / 5
Completeness	An explicit 'Use when' clause answers 'when', and 'executing implementation plans' weakly implies the 'what', but the actual behavior (fresh subagent per task + reviews) is never stated, so the 'what' is only implied within the trigger clause.	3 / 5
Trigger Term Quality	'executing implementation plans', 'independent tasks', and 'in the current session' are relevant trigger terms, but common variations a user might say (e.g. 'run this plan', 'execute tasks') are missing.	3 / 5
Distinctiveness Conflict Risk	'independent tasks' and 'in the current session' carve a clear niche distinct from the sibling executing-plans skill, with only minor overlap risk against that closely related skill.	4 / 5
	Total	12 / 20 Passed

Validation

93%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
relative_links	Relative link issues: 2 missing, 1 suspicious	Warning

	Total	15 / 16 Passed

Repository: rpamis/comet
Path: eval/local/skills/benchmarks/dependency/superpowers/subagent-driven-development/SKILL.md
Commit: f329b48

Reviewed: 1 day ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.