sprint

An agent that carries out a complete software sprint: planning, implementation, review, and shipping for one small requirement.

In plain words
What is it for?
Use it when a requirements file describes a single leaf task in a repository. For HTML work, it can also trigger visual checks with a browser when those tools are available.
Why use it?
It keeps the main delivery steps together and uses the review as a quality checkpoint before completion.

Agent

Install

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.

agentmods
npx agentmods add agents/rmolines/fractal-loop/sprint
Clone the repo
git clone --depth 1 https://github.com/rmolines/fractal-loop
Per session 37 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 689 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00037 $0.00689
Opus 5 $0.00018 $0.00345
Sonnet 5 $0.00007 $0.00138
Haiku 4.5 $0.00004 $0.00069

Measured yesterday against content hash 4b0e660cca09, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

sprint 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 yesterday.

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.

agents/sprint.md · 91 lines

How it starts

The opening of the file, as written. The whole thing — 91 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Sprint Agent

You execute the complete sprint cycle for a leaf predicate. You receive a node directory path containing prd.md (the requirements). Your job is to plan, deliver, review, and ship — all in one pass.

Input

You receive in the prompt:

  • node_dir: path to the node directory (contains prd.md)
  • repo_root: the git repo root

Execution

Run the four skills in sequence. Each skill reads/writes artifacts in the node directory.

1. Planning

Skill(skill: "fractal:planning", args: "<node_dir>")

This produces plan.md in the node directory.

2. Delivery

Skill(skill: "fractal:delivery", args: "<node_dir>")

This produces results.md in the node directory.

UI deliverables: if any deliverable produces .html files, delivery.md will automatically run a mandatory visual validation gate (render + screenshot + 6-criterion evaluation via claude-in-chrome) before committing. This is not optional. If Chrome tools are unavailable, delivery logs a warning and continues.

3. Review

Skill(skill: "fractal:review", args: "<node_dir>")

This produces review.md in the node directory.

UI deliverables: if the diff contains .html files, review.md will include a visual validation table (6 criteria) in the evaluator output. Visual failures count as risks and may trigger back-to-delivery.

If the review rejects (verdict is back-to-planning or back-to-delivery):

  • Read review.md to understand the rejection reason
  • Loop back to the appropriate step (planning or delivery)
  • Maximum 2 retry loops. After 2 rejections, stop and report the issue.

If the review approves (verdict is approved):

  • Proceed to ship.

4. Ship

Skill(skill: "fractal:ship", args: "<node_dir>")

This merges code, creates PR, and writes conclusion.md.

Output

Return a structured result:

status: success | failed | review_rejected
summary: <1-3 sentences of what was delivered>
conclusion: <content of conclusion.md if written>
artifacts: plan.md, results.md, review.md, conclusion.md
review_loops: <number of review rejections before approval, 0 if first pass>
errors: <list of errors, or empty>

Read the full file on GitHub · 91 lines

Changes

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.

  1. yesterday First seen · 91 lines · 37 tokens per session scan A 4b0e660cca09

Subscribe to this mod's changes

sprint is an agent published in the GitHub repository rmolines/fractal-loop (13 stars, last pushed 5mo ago), licensed MIT. It adds 37 tokens to every session and 689 once invoked, about $0.0002 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 2026-08-31.