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 a1f/agent-templates --skill make-prgit clone --depth 1 https://github.com/a1f/agent-templatesWrote 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/a1f/agent-templates/make-pr)<a href="https://agentmods.dev/skills/a1f/agent-templates/make-pr"><img src="https://agentmods.dev/badge/skills/a1f/agent-templates/make-pr/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/a1f/agent-templates/make-pr"><img src="https://agentmods.dev/badge/skills/a1f/agent-templates/make-pr.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 | $0.00101 | $0.05378 |
| Opus 5.5 | $0.00040 | $0.02151 |
| Sonnet 5.5 | $0.00020 | $0.01076 |
| Haiku 4.5 | $0.00010 | $0.00538 |
Grade A, and why
make-pr 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 2mo 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 — 300 lines — stays where its author put it; the contents beside it link to each section on GitHub.
make-pr
You are the architect: the workflow that takes one already-scoped task (one PR, one module) and drives it to done — dispatching worker agents in a fixed order, running the gates, and logging every subagent call so a human can validate the run afterward. You never write the production code or tests yourself: you verify the agents' structured returns and decide the next step.
Scope boundary: the task arrives scoped to a single module/PR and must name the module boundary (
module,allowed_paths, or equivalent) and the public interface or files in scope. If it spans more than one module or the boundary is missing, stop and report for re-scope — do not proceed.
Agents
| Agent | When | Job |
|---|---|---|
tdd-runner |
RED, once per behavior | write ONE failing test, prove it fails for the right reason |
worker-coder |
GREEN & REFACTOR (per behavior); non-behavioral edits | minimal code to pass the test, a behavior-preserving refactor, or an exact non-behavioral change; commit |
reviewer |
after gates are green | quality + bugs + security + line-by-line rule conformance on the diff |
comment-reviewer |
with the reviewer, on the gate-green diff | comment quality: scores the added comments against comments.md, each finding carrying its replacement text |
critic |
after both reviewers, on the gate-green diff | goal-fit: did the PR achieve the task? |
Dispatch via the Agent tool. Give each agent the task context + base + target_cwd +
module boundary + allowed paths + exact files it needs + absolute rule paths, and give
worker-coder its mode (green | refactor | non_behavioral) plus
dependencies_allowed (and any named dependency/version) when the task permits dependencies —
otherwise the coder blocks on any new dependency. Keep each dispatch prompt tight and scoped
to one job. Require each agent to return exactly the JSON object defined in its prompt, and
validate that return against ~/.claude/at/schemas/<role>.schema.json (keyed by the
return's role field). The validator reads the instance from a file: write the raw
return verbatim with Write (not echo/heredoc — those mangle embedded quotes and
newlines) to <run_root>/<run-id>.<role>.json, then run:
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.
- 2mo ago First seen · 300 lines · 0 tokens per session scan A e6f957d5996f
make-pr is a skill published in the GitHub repository a1f/agent-templates (2 stars, last pushed 2mo ago), licensed MIT. It adds 101 tokens to every session and 5,378 once invoked, about $0.0004 per session on Opus 5.5. 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 2026-08-31.
Other skills, from other repositories
workflow-patterns
Use this skill when implementing tasks according to Conductor's TDD workflow, handling phase checkpoints, managing git commits for tasks, or understanding the verification protocol.
cmux-testing
Choose scoped cmux verification, add behavioral tests, and validate Swift test targets and wiring. Use when adding tests or deciding what local/CI evidence a change needs.
nw-quality-framework
Quality gates - 11 commit readiness gates, build/test protocol, validation checkpoints, and quality metrics.
strict-tdd
Strict RED->GREEN->REFACTOR test-driven development with enforcement. Never write production code before a failing test. Atomic commits per TDD cycle.
loom-builder-pr
This document covers PR creation, test output handling, and quality requirements for the Builder role. For the core builder workflow, see builder.md.
resolve-issues
Resolves GitHub issues using isolated worktrees and test-driven development, then delegates PR creation to /github:create-pr so the quality gate and the /github:review-pr loop always run. This skill should be used when the user asks to "resolve an issue", "fix issue.