BuildNot yet confirmed elsewhere1 publisher3 min readPublished
Superpowers makes spec-driven work a precondition, then ships it to twelve harnesses
The enforcement is a bootstrap prompt the agent checks before every task, so its durability depends on per-harness plumbing. On Hermes Agent, compaction can lose it.
The Engineer · Build desk

What happened
- Superpowers ships a spec-driven methodology as installable skills, with brainstorming, planning, subagent review and red-green-refactor TDD as steps the agent checks before every task.
- The dev.to quickstart frames it against a known failure: a hand-rolled specify-plan-implement skill gets used diligently for a week, then dropped when a deadline closes.
- Plugin manifests cover twelve harnesses, among them Claude Code, Cursor, Codex, Gemini CLI, Kimi Code, Devin CLI and Factory Droid.
Why it matters
- constraint The rules only bind while the bootstrap is in context, so on Hermes Agent it is compaction, not the operator, that decides whether TDD is still mandatory mid-session.
- cost Twelve manifests mean twelve install and update paths, and OpenCode's separate path on a shared machine turns team consistency into per-seat, per-tool work.
- capability Changing coding agents no longer resets your process, because the workflow is a distributable package rather than a folder in one repository.
- precedent An opinionated methodology carried in a vendor plugin marketplace makes installing the process, not just the tool, a normal part of setup.
The enforcement mechanism is a prompt. The differentiator, by the dev.to quickstart's own account, is a bootstrap instruction that makes the agent check for relevant skills before it starts any task, so the workflow activates without anyone typing a slash command [3]. That is a real improvement on a skill file whose invocation depends on a human remembering it while a deadline closes [20]. It is also a guarantee that lives in the context window, and the install notes say so plainly: Hermes Agent has no post-compaction hook, so a long session that compacts over its first turn can lose the bootstrap, and the documented remedy is to start a fresh session [9]. Pi approaches the same risk from the other side, with a small extension that injects the `using-superpowers` bootstrap at session start and again after compaction, and which does not need Pi's compatibility Skill tool because Pi supports skills natively [8]. One methodology, two different odds of it being in force on turn forty.
Then the portability accounting. Twelve harnesses have manifests [4][17], and there is no global install; you install per agent [5]. OpenCode counts as a separate path even on a machine where another harness is already set up [10], Claude Code has two marketplaces to pick between, Anthropic's official one and obra/superpowers-marketplace [6], and Cursor users find it by searching the plugin UI [16]. Portability means the methodology survives a tool switch, not that one install covers your desk.
Worth noting from the same install section: Antigravity gets its own note, including that it runs the session-start hook automatically so the skills are live from the first message [7], but Antigravity does not appear in the twelve-name manifest list [18]. Of the twelve that are listed, three carry harness-specific caveats [19]. The supported surface is already moving faster than the list that describes it.
The success signal is conversational. The guide's advice is to ask the agent what skills are available rather than trust a quiet install message, and to look for `brainstorming`, `writing-plans`, `test-driven-development` and `subagent-driven-development` in the reply [11]. When a skill is present but never fires, the guide puts the fault in the routing description rather than the model [12]. So the debugging surface is prose, and nothing here returns an exit code you could wire into CI.
What the package actually relocates is where the discipline is stored: out of the operator's memory and into the harness startup sequence [3]. Seven skills are documented as the basic workflow [15], and "write tests first, always" is one of four stated principles the agent is instructed to check [13]. That is a better home for the rule than a README, and it is still not a lock.
What to watch
- Whether Hermes Agent gains a post-compaction hook, closing the one documented place where the workflow can go quiet mid-session.
- Whether updating stays a manual reinstall on each harness or a marketplace refresh handles it, since twelve installs multiply either way.
- Any published evidence, rather than assertion, that teams stay on the workflow past the week where hand-rolled skill files get abandoned.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence34
- Adoption
- Insufficient
- Hype gap+30
- Incentives
- Insufficient
- Confidence41
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
Superpowers packages a full spec-driven methodology into installable Claude Skills, enforcing brainstorming, planning, subagent-driven implementation and strict TDD instead of leaving that structure up to the user.
- [2]
Superpowers is a skills package built by Jesse Vincent and the team at Prime Radiant that treats brainstorming, planning, subagent review and red-green-refactor TDD as mandatory steps the agent checks before every task rather than optional suggestions.
- [3]
The stated differentiator is a bootstrap instruction that forces the agent to check for relevant skills before starting any task, so the workflow activates automatically instead of requiring the user to remember a slash command.
- [4]
Superpowers ships plugin manifests for Claude Code, Cursor, Codex, Gemini CLI, GitHub Copilot CLI, Grok Build CLI, Kimi Code, OpenCode, Pi, Devin CLI, Factory Droid and Hermes Agent, so the same methodology follows the user across harnesses instead of living in one repo's .claude/skills/ folder.
- [5]
Installation is harness-specific and the project is explicit that it must be installed separately for each agent used; there is no single global install.
- [6]
For Claude Code, Superpowers is distributed through two marketplaces: the official Anthropic plugin marketplace and the dedicated obra/superpowers-marketplace, which also carries a few related plugins.
- [7]
Antigravity runs the plugin's session-start hook automatically, so Superpowers is active from the first message, and reinstalling with the same command updates it.
- [8]
Pi loads the skills through a small extension that injects the using-superpowers bootstrap at session startup and after context compaction, and does not require Pi's compatibility Skill tool because Pi has native skill support.
- [9]
Hermes Agent has no post-compaction hook, so a very long session that compacts over its first turn can lose the bootstrap; the guide advises starting a fresh session if skills stop triggering.
- [10]
OpenCode treats Superpowers as a completely separate install path from any other harness already in use, even on the same machine.
- [11]
The guide advises verifying the install by asking the agent "What skills are available?" rather than trusting a silent success message; a correct install should list skills such as brainstorming, writing-plans, test-driven-development and subagent-driven-development.
- [12]
If a skill exists but never activates, the guide says the routing description is almost always the problem, not the model.
- [13]
The project's stated philosophy has four points: test-driven development, write tests first always; systematic over ad-hoc; complexity reduction as the primary goal; and evidence over claims, verify before declaring success.
- [14]
Superpowers describes itself as "a complete software development methodology for your coding agents, built on top of a set of composable skills and some initial instructions that make sure your agent uses them."
- [15]
Superpowers documents seven skills as "the basic workflow," each activating at a specific point.
- [16]
Superpowers can also be found by searching "superpowers" directly in Cursor's plugin marketplace UI.
- [17]
The manifest list names twelve harnesses.
- [18]
Antigravity is covered by the guide's install notes but does not appear in the twelve-harness manifest list.
- [19]
Three of the twelve harnesses with manifests carry documented harness-specific install or bootstrap caveats: Pi, Hermes Agent and OpenCode.
- [20]
The dev.to guide describes the common failure mode of Claude Code SDD setups: someone writes a solid specify-plan-implement skill once, uses it diligently for a week, then quietly drifts back to unstructured prompting the first time a deadline gets close, and nothing in a hand-rolled skill stops that.
Sources
1 independent publisher whose own reporting we read for this story.
- dev.toSuperpowers Quickstart: Install, Workflow, and Tryout
1 article · August 24, 2026
Topics and entities
Follow any of these and your For You feed starts watching them — no settings page required.
Topics
- Cross-Harness Tooling PortabilityFollow
- AI Coding AgentsFollow
- Agent Skills PackagingFollow
- Spec-Driven DevelopmentFollow
- TDD Enforcement in Agent WorkflowsFollow
- Context Compaction and Bootstrap ReliabilityFollow
Entities
- superpowersFollow
- Jesse VincentFollow
- Prime RadiantFollow
- Claude CodeFollow
- AnthropicFollow
- obra/superpowers-marketplaceFollow
- CursorFollow
- OpenCodeFollow
- Pi AgentFollow
- Hermes AgentFollow
- AntigravityFollow
- Codex CLIFollow
- Gemini CLIFollow
- CopilotFollow
- Grok Build CLIFollow
- Kimi CodeFollow
- Devin CLIFollow
- Factory DroidFollow