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 skills add brandhaug/b2b-saas-starter --skill implementgit clone --depth 1 https://github.com/brandhaug/b2b-saas-starterWrote 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/brandhaug/b2b-saas-starter/implement)<a href="https://agentmods.dev/skills/brandhaug/b2b-saas-starter/implement"><img src="https://agentmods.dev/badge/skills/brandhaug/b2b-saas-starter/implement/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/brandhaug/b2b-saas-starter/implement"><img src="https://agentmods.dev/badge/skills/brandhaug/b2b-saas-starter/implement.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00022 | $0.01042 |
| Opus 5 | $0.00011 | $0.00521 |
| Sonnet 5 | $0.00004 | $0.00208 |
| Haiku 4.5 | $0.00002 | $0.00104 |
Grade A, and why
implement 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 — 75 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Implement
Default endpoint: a reviewed PR ready to merge, without merging. Honor a narrower user request.
-
Read the issue and comments using tracker guidance. Keep issue numbers and acceptance-criteria (AC) IDs only in GitHub issue/PR traceability. Use descriptive behavioral names in repository code, docs, comments, test titles, filenames, and other tracked implementation content; never embed issue numbers or AC IDs there. Preserve legitimate external dependency references and unrelated numeric values. Map criteria to test evidence in the PR. Use
grillingfor unresolved product decisions andto-ticketsfor independently verifiable slices. -
Use the host's available delegation tools and configured model IDs. Keep the session's provider unless the user requests a switch. For bounded implementation, review, and browser work, prefer GLM 5.3 → GLM 5.3 Flash, Claude Opus → Sonnet, Codex Astra → Luna. Resolve these pairings through the host's configured model IDs or worker profiles and select the worker model explicitly. For other models, use a configured cheaper same-provider model. Inherit the session model only after checking that no suitable worker model is available, and report the fallback. Keep coordination and final decisions in the main session. If delegation is unavailable, work locally and report that review was not independent. Give workers compact briefs with ACs, owned files, and relevant pointers. Start with fresh context where supported; include conversation history only when the task depends on it. Keep tiny tasks local. Default to one implementation worker; add workers for independent file boundaries. Assign one writer to shared wiring. The coordinator alone dispatches reviews and includes this ownership rule in worker briefs. Implementers and reviewers do not dispatch additional reviewers.
Use the context controls exposed by the active tool schema:
- Codex: set
fork_turns: "none"when exposed; its default can inherit the full conversation. If the tool exposesfork_contextinstead, set it tofalse. - Claude Code: start a new
Agentwithsubagent_type: "general-purpose"or a configured non-fork subagent. These start fresh;forkcopies history. - OpenCode: start a new
taskwith a compactpromptand omittask_id. Supplyingtask_idresumes an existing worker's context.
Resume an existing worker when continuing its assigned task.
- Codex: set
-
Apply
effectwith repo examples,codebase-designfor module interfaces,effect-service-designfor service ownership/Layers, andimpeccablefor UI. Usetddat agreed behavior seams. If a supporting skill is missing, follow skills setup. -
Docs and mechanical edits get one targeted reviewer. For behavior changes, use
code-reviewagainst a fixed base/head and the agreed spec, with its separate Standards and Spec reviewers. The Standards reviewer appliesthermo-nuclear-code-quality-reviewin the same pass and deduplicates overlapping findings. Keep Spec independent; run a separate deep audit only on explicit request. -
Consolidate findings before assigning fixes. Structural suggestions need a concrete benefit. Send accepted repairs back to the original implementer. Re-review only accepted findings plus the repair diff and affected callers. If two repair rounds fail, reassess the cause. Supply missing context or use a fresh worker on a more capable configured same-provider model, carrying prior attempts and open findings. If that repair still fails, report the unresolved blockers and stop repair dispatch. Repeat the broader review only when design or behavior changes substantially.
-
For UI acceptance gaps, use available browser tooling to run one smoke path. Give the verifier the URL, credential source, actions, expected visible outcomes, and a limit of 12 browser actions. Stop on the first failure and return pass/fail evidence; gather screenshots or console/network diagnostics as needed. Prefer existing Playwright coverage for regression checks. If browser tooling is unavailable, report the unverified acceptance criteria.
-
Run focused checks during work and
pnpm run validateon the integrated result. Commit and create/update the PR using the repo template. -
Run
gh pr checks --required --watchwith a 30-minute deadline per head. Fix actionable failures and reverify after pushes. Before finishing, confirm the same head satisfies required checks/reviews and is up to date with and mergeable against the base. Report optional failures and external blockers.
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 · +5 lines 4e6bec25c1d6
- 4d ago First seen · 70 lines · 22 tokens per session scan A c32aa6fe9fb8
implement is a skill published in the GitHub repository brandhaug/b2b-saas-starter (42 stars, last pushed yesterday), licensed MIT. It adds 22 tokens to every session and 1,042 once invoked, about $0.0001 per session on Opus 5. 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 2026-09-07.
Other skills, from other repositories
add-a-lib
Scaffold a new shared library under libs/ in the builders-stack monorepo. Use when code is needed in two or more places (apps or services) and should become a single source of truth consumed by package name. Covers the package.json, tsconfig, the one-public-door src/index.ts barrel, and wiring it into a consumer…
project-setup-info-local
Comprehensive setup steps to help the user create complete project structures in a VS Code workspace; this tool is designed for full project initialization and scaffolding, not for creating individual files. When to use this tool: user wants to create a new complete project from scratch; setting up entire project…
turborepo
Turborepo monorepo build system guidance. Triggers on: turbo.json, task pipelines, dependsOn, caching, remote cache, the "turbo" CLI, --filter, --affected, CI optimization, environment variables, internal packages, monorepo structure/best practices, and boundaries. Use when user: configures tasks/workflows/pipelines…
project-setup-info-context7
Comprehensive setup steps to help the user create complete project structures in a VS Code workspace. This tool is designed for full project initialization and scaffolding, not for creating individual files. When to use this tool: when the user wants to create a new complete project from scratch; when setting up…
triage
Classify an issue, identify missing information, and decide whether bounded automatic work is safe.
record-gotchas
Capture surprises, workarounds, and rough edges hit while consuming the public surface of Prisma 8, Prisma Compute, or Prisma Postgres — anything a real user of these products would experience. Fires whenever an operator (or agent) writes a workaround, hits a surprising failure mode, or finds undocumented behaviour…