save

A checkpoint command that records the current project state in planning files. It saves what is finished, in progress, next, blocked, and changed so a later session can continue.

In plain words
What is it for?
Use it before closing a session or after a major milestone to save project progress for the next session.
Why use it?
Coding-agent sessions can end before the work is complete, leaving context behind. A written checkpoint makes the handoff clear without undoing progress.

Command

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 commands/randomittin/heimdall/save
Clone the repo
git clone --depth 1 https://github.com/randomittin/heimdall
Per session 55 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 1,598 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 $0.00055 $0.01598
Opus 5 $0.00028 $0.00799
Sonnet 5 $0.00011 $0.00320
Haiku 4.5 $0.00006 $0.00160

Measured yesterday against content hash 3201f0ce2b46, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

save 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.

commands/save.md · 138 lines

How it starts

The opening of the file, as written. The whole thing — 138 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Save — Checkpoint Current State

Create or update ALL .planning/ state files so the next heimdall run resumes exactly where you left off.

What to save

Read the current state of the project and write/update these files:

1. .planning/STATE.md

Update with:

  • Current phase: what phase are we in (planning, executing wave N, verifying, idle)
  • What's done: list completed tasks/features with commit hashes
  • What's in progress: currently active work
  • What's next: queued tasks not yet started
  • Blockers: anything preventing progress
  • Decisions made: key architectural/design decisions from this session
  • Key files changed: list of files modified in this session

2. .planning/CHECKPOINT.md (the handoff note — most important file)

Write a concise handoff note that another Claude session can read and immediately continue:

# Checkpoint — [timestamp]

## TL;DR
[One sentence: what was the task, how far did we get]

## Completed
- [x] Task 1 (commit abc1234)
- [x] Task 2 (commit def5678)

## In Progress
- [ ] Task 3 — started, file X partially edited

## Not Started
- [ ] Task 4
- [ ] Task 5

## Resume Instructions
[Exact next step: "Run tests in src/auth/, fix any failures, then proceed to Task 4"]

## Key Context
- [Decision 1: chose X over Y because Z]
- [File A imports from File B — don't break this]
- [User wants: specific preference]

## Tech Stack & Patterns
- [Stack: React 18 + TypeScript + Vite, etc.]
- [Pattern: services use singleton pattern, components use compound pattern]
- [Constraint: must support Node 18+, no ESM-only deps]

## Project Settings
- Parallelism: [max 10 / max 5 / sequential — what worked for this project]
- Model routing: [any overrides from defaults, e.g. "sonnet works fine for React components here"]
- Governance: [hierarchical / democratic / emergency last used]
- Test command: [exact command, e.g. "npm test", "pytest tests/", "cargo test"]
- Lint command: [exact command, e.g. "npx eslint src/", "ruff check ."]
- Build command: [exact command, e.g. "npm run build", "cargo build"]
- Deploy command: [if known]
- Directories to avoid: [e.g. "vendor/, generated/, dist/"]
- User preferences: [e.g. "prefers tabs over spaces", "wants detailed commit messages", "hates emojis in code"]

Read the full file on GitHub · 138 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 · 138 lines · 55 tokens per session scan A 3201f0ce2b46

Subscribe to this mod's changes

save is a command published in the GitHub repository randomittin/heimdall (5 stars, last pushed 11d ago), licensed MIT. It adds 55 tokens to every session and 1,598 once invoked, about $0.0003 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.