Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/will-pagane/claude-superdev-harness/session-buildnpx skills add will-pagane/claude-superdev-harness --skill session-buildgit clone --depth 1 https://github.com/will-pagane/claude-superdev-harnessWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/will-pagane/claude-superdev-harness/session-build)<a href="https://agentmods.dev/skills/will-pagane/claude-superdev-harness/session-build"><img src="https://agentmods.dev/badge/skills/will-pagane/claude-superdev-harness/session-build.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00000 | $0.04013 |
| Opus 5 | $0.00000 | $0.02006 |
| Sonnet 5 | $0.00000 | $0.00803 |
| Haiku 4.5 | $0.00000 | $0.00401 |
Grade A, and why
session-build scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 161 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Session Build
Idea in → one or more verified, pushed branches out, in one pass.
Pipeline: brainstorm → spec(s) → plan → codex-review → implementation → verified push.
Migrations and deploys are tasks inside the plan, not a separate phase.
Pull request and delivery merge are not part of this skill at all — /session-end owns them, run by the user after their own verification. (The one merge this skill does perform is a fork taking a peer's branch on the orchestrator's explicit order; see invariant 1.)
Announce at start: "Using session-build to take this idea from brainstorm to pushed branch(es)."
This file is a router. It holds what is true for every run; each step file holds its own work and names its successor. Read one step file at a time, all the way through, and never load a later step before the current one is complete.
Entry points
| Invocation | Start at |
|---|---|
/session-build <prompt> |
steps/step-01-preflight-and-brainstorm.md — the prompt seeds the brainstorm |
/session-build, specs already written this session |
steps/step-02-scope-and-collisions.md — confirm which specs are in scope, skip the brainstorm |
/session-build, no prompt and no specs |
Ask for the idea in one message, then step-01 |
| The prompt carries an already-written spec or plan | step-02. Do not invent a ruling about whether the brainstorm still applies — say in one line that you are entering at scope confirmation because the spec already exists, and re-validate its premises against the live tree (step-01 owns that check; run it even though you skip the rest of step-01). A plan written in an earlier session was hardened against the code as it was then. |
| You were launched as a fork by an orchestrator | Read references/fork-contract.md, then steps/step-04-build.md. Nothing else. Steps 1, 2, 5 and 6 belong to the orchestrator; reading them spends your context on work you are forbidden to do. |
| Mid-conversation drift — this conversation already produced a spec, a plan, or a code edit toward a multi-part build, and this skill was never invoked | Stop. Say so in one line. Then step-01, treating everything already produced as input to be re-validated rather than progress to be preserved. The work so far happened in the shared checkout with no isolation; step-03 is what fixes that. |
| Conversational invocation — the user explained the idea over several turns and then asked for a build in their own words ("builda isso", "faça as melhorias", "roda o session build") rather than typing the slash command | step-01, at full strength. This is the entry point that gets degraded, and the degradation is always the same: the conversation already feels like a brainstorm, so superpowers:brainstorming is skipped, and from there the rest of the chain falls with it. A conversation is not a brainstorm. It has produced no spec file, cleared no design gate, and had no premise checked against the live tree. Announce the chain (below) and run every link. |
What ships with it
18 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- assets/closeout-report-template.md 3.9 KB
- dispatch-prompts.md 23 KB
- references/collision-precedents.md 4.4 KB
- references/fork-contract.md 11 KB
- references/incidents.md 9.9 KB
- references/measurement-discipline.md 6.2 KB
- references/orchestration-deadlocks.md 24 KB
- references/verification-epistemology.md 11 KB
- scripts/gate.ps1 3.5 KB runs code
- scripts/gate.sh 7.0 KB runs code
- scripts/ledger.py 17 KB runs code
- scripts/preflight.sh 5.3 KB runs code
- steps/step-01-preflight-and-brainstorm.md 7.1 KB
- steps/step-02-scope-and-collisions.md 11 KB
- steps/step-03-isolate.md 7.1 KB
- steps/step-04-build.md 12 KB
- steps/step-05-orchestrate.md 19 KB
- steps/step-06-closeout.md 8.7 KB
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago Changed · +2 lines 79857bcaf2a2
- 6d ago First seen · 159 lines · 0 tokens per session scan A bca13937aa3c
session-build is a skill published in the GitHub repository will-pagane/claude-superdev-harness (4 stars, last pushed 3d ago), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 4,013 tokens. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
aenv
Use when the user wants to set up, switch between, or manage aenv namespaces — named bundles of CLAUDE.md, skills, MCP entries, and other AI-coding-harness config — in a project OR globally across $HOME. Triggers include aenv … mentioned directly, "switch namespace/profile", "activate/deactivate", "create/snapshot a…
product-analytics
Product analytics and growth expert. Use when designing event tracking, defining metrics, running A/B tests, or analyzing retention. Covers AARRR framework, funnel analysis, cohort analysis, and experimentation.
api-design
REST/GraphQL/gRPC API design best practices. Use when designing APIs, defining contracts, handling versioning. Covers OpenAPI 3.2, GraphQL Federation, gRPC streaming.
merge-pr
Merge a pull request with status checks, squash merge, and branch cleanup. Handles worktree contexts.
at-vision
Inspect screenshots, photos, diagrams, image paths, and image URLs when the task depends on visible content. Use when the prompt lacks actual image content, native inspection fails, or the user requests inspectimage; prefer the MCP tool, then the installed CLI.
general-video
The fallback workflow for authoring custom HyperFrames video compositions at any length or format — longer or multi-scene pieces, brand / sizzle reels, montages, title cards, static loops, and freeform compositions. Input- and length-agnostic. If a specialized workflow clearly fits the input — a marketed product, a…