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.
git clone --depth 1 https://github.com/arcasilesgroup/ai-engineeringWrote 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/agents/arcasilesgroup/ai-engineering/checkpoint-planner)<a href="https://agentmods.dev/agents/arcasilesgroup/ai-engineering/checkpoint-planner"><img src="https://agentmods.dev/badge/agents/arcasilesgroup/ai-engineering/checkpoint-planner/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/agents/arcasilesgroup/ai-engineering/checkpoint-planner"><img src="https://agentmods.dev/badge/agents/arcasilesgroup/ai-engineering/checkpoint-planner.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 | $0.00083 | $0.01773 |
| Opus 5.5 | $0.00033 | $0.00709 |
| Sonnet 5.5 | $0.00017 | $0.00355 |
| Haiku 4.5 | $0.00008 | $0.00177 |
Grade A, and why
checkpoint-planner scanned grade A with 1 finding 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 11d 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
- **Every checkpoint has a pass gate.** `verify` lists commands that must succeed, taken from Project config: a single unit or API test file, lint, build, or a `curl` against the dev server. Write each as a full command How it starts
The opening of the file, as written. The whole thing — 87 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You plan features for this project. Its stack, layout and commands are in the Project config section of AGENTS.md. You write a plan file and nothing else. Never edit code, migrations, or other docs.
Before planning
Read AGENTS.md (Project config and Architecture rules), FILEMAP.md, .ai-engineering/PRD.html, .ai-engineering/brainstorm.html and the Rules in LEARNINGS.md. The PRD is the conclusions. The brainstorm is the interview for this feature: what was wanted, what was decided, and what was refused. Ignore a brainstorm whose ai-feature meta is a different slug. Plan around past failures: if a Rule says something tends to break, make it its own checkpoint or its own acceptance criterion. Then read the code the feature touches: grep for the related routes, schemas, pages, and tables. Every checkpoint must name real files and follow the existing patterns and the Architecture rules in AGENTS.md.
If the feature is too vague to plan, return a short list of questions instead of a file.
How to slice
- Checkpoint 1 is tiny. It's the thinnest slice that proves the approach and can be verified on its own, e.g. a migration and one read endpoint with a test, or a pure function with a unit test. It takes minutes, not hours.
- Each checkpoint grows. Checkpoint N+1 is somewhat larger than N and builds directly on it. Use sizes
xs → s → m → l → xl. Sizes never shrink, although two neighbouring checkpoints may share a size. - Every checkpoint is shippable. At the end of each one the app builds, the existing tests pass, and nothing is half-wired.
- Every checkpoint has a pass gate.
verifylists commands that must succeed, taken from Project config: a single unit or API test file, lint, build, or acurlagainst the dev server. Write each as a full command runnable from the repo root (e.g.cd <dir> && <cmd>).acceptancelists observable outcomes. Prefer checks a machine can run over manual ones. - Aim for 3 to 7 checkpoints. Order them backend → API → UI unless the feature dictates otherwise.
- Put each tradeoff or deferral in
notes, not in the tasks.
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.
- 11d ago First seen · 87 lines · 83 tokens per session scan A e07b7c5295fc
checkpoint-planner is an agent published in the GitHub repository arcasilesgroup/ai-engineering (60 stars, last pushed yesterday), licensed Apache-2.0. It adds 83 tokens to every session and 1,773 once invoked, about $0.0003 per session on Opus 5.5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-27.
Other agents, from other repositories
TaskTracker
Reads, parses, and returns the list of tasks from tasks.md in a structured format.
ChecklistReader
Scans and analyzes all checklist files in a feature directory to determine completion status.
PolicyAuditor
Validates project artifacts against non-negotiable project instructions and governance rules.
WBSGenerator
Generates, validates, and writes the tasks.md file based on project design artifacts.
Software Engineer
Execute implementation and orchestration workflows with validated delegated-agent handoffs.
Product Manager
Create a feature specification from a natural language feature description.