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/dsswift/ionWrote 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/commands/dsswift/ion/align)<a href="https://agentmods.dev/commands/dsswift/ion/align"><img src="https://agentmods.dev/badge/commands/dsswift/ion/align/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/commands/dsswift/ion/align"><img src="https://agentmods.dev/badge/commands/dsswift/ion/align.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 | $0.00051 | $0.08013 |
| Opus 5.5 | $0.00020 | $0.03205 |
| Sonnet 5.5 | $0.00010 | $0.01603 |
| Haiku 4.5 | $0.00005 | $0.00801 |
Grade A, and why
align 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 16d 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 — 658 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/align
ARGS means the raw arguments of this invocation:
$ARGUMENTS
An empty block means no arguments were passed.
/align has two modes.
| Mode | Runs when | Reviews | Writes |
|---|---|---|---|
| A: Plan Alignment | The conversation is in plan mode | The plan, before any code exists | The plan file only |
| B: Post-Changes Alignment | Everything else | A diff: the local branch, a named branch, or a pull request | A fix plan, then after approval the fixes and their commits |
Limits
These hold in both modes.
- Review the whole target. Mode and ARGS set the review surface. Size never does. A large diff produces a large report. The only narrowing is a focus or a target the operator typed in ARGS, so there is never a question to ask about scope.
- Review content, not commit shape or PR lifecycle.
/squashowns commit partitioning./create-prowns the pull request. The operator invokes both. No finding, amendment, recommendation, plan step, or open item sends the operator to either one, asks for a merge strategy, or asks for commits to be split, reordered, or squashed. - No remote writes. No
git push, nogh pr create,gh pr merge,gh pr review, orgh pr comment. - No history reshaping. Commit count, order, and scope seams stay as they are. The one permitted rewrite is the amend delivery in B-Step 6.
- No commit before approval. Mode A never commits. Mode B commits only after the operator approves its fix plan.
- One report. The analysis is a single markdown report in this chat. The plan write follows it.
Pick the mode
Decide once, when the command fires. Use the first row that matches.
| ARGS or state | Mode |
|---|---|
ARGS is only PR references, or starts with in PR |
B, PR mode |
ARGS starts with in branch |
B, Branch mode |
| Plan mode is active | A |
| Anything else | B, Local mode |
A PR reference is 287, #287, PR 287, PR #287, or a list of these separated by commas or spaces. A PR or branch target wins over plan mode and over the state of the local checkout, because the operator named what to review.
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.
- 16d ago Changed · -371 lines · +12 tokens per session 77a964cca9ed
- 1mo ago Changed · -1 lines aeae5048fd62
- 1mo ago First seen · 1,030 lines · 39 tokens per session scan A f5b020af6612
align is a command published in the GitHub repository dsswift/ion (4 stars, last pushed yesterday), licensed MIT. It adds 51 tokens to every session and 8,013 once invoked, about $0.0002 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 commands, from other repositories
undo-commit
Soft-undo the last commit. Restores its changes as staged, leaves working tree intact. Refuses if HEAD is already pushed to a tracked remote.
session-report
Capture what changed this session and why, scoped to the current branch. Read by ship verbs when synthesizing the commit message; deleted after a successful commit.
commit
Stage, commit, and optionally ship further. Pass an escalation token (push, pr, merge, squash, squash merge) to skip the prompt. With no token, commits then asks how far to ship. Delegates message format to the atomic-git-discipline skill.
hatch3r-pr-resolve
Read open PR comments, evaluate each against current code via the rigor contract, implement accepted findings, reply inline. Multi-platform.
afrv
Revise the current feature worktree after code review — decide accept/revert/modify (shortcut for feature-code-revise).
implement-revalidate
Read by /implement when the user asks to re-validate (D-CMD-REF-HAND).