automode-edit

automode-edit is a command for Claude Code from obeone/claude-skills. It costs 43 tokens per session (2,335 once invoked), scanned B, original, MIT.

A plain-language editor for the autoMode section of .claude/settings.local.json. It turns a requested rule change into a checked proposal and applies it through validation and an atomic file write.

In plain words
What is it for?
Use it to add, remove, rename, or restore natural-language environment, allow, soft-deny, and hard-deny rules, such as controlling deploys or pushes to main.
Why use it?
It reduces the risk of manually editing security-related automation rules incorrectly or silently changing their meaning.

Command for Claude Code

Written for Claude Code: a Claude Code command (commands/*.md). Also seen: reads .claude/ paths.

Not installable on its own: it runs a file from its repository that does not travel with it. Clone the repository, or install whatever ships that file. The line is uv run skills/automode-config/scripts/inspect_automode.py --json --file all.

Install

Getting it into your agent

There is no command for this one: it runs only inside a plugin, and the catalogue could not identify which plugin ships it. The source is linked below.

Made for: Claude Code.

Wrote 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.

agentmods badge for automode-edit

README.md
[![agentmods](https://agentmods.dev/badge/commands/obeone/claude-skills/automode-edit.svg)](https://agentmods.dev/commands/obeone/claude-skills/automode-edit)
Your own site
<a href="https://agentmods.dev/commands/obeone/claude-skills/automode-edit"><img src="https://agentmods.dev/badge/commands/obeone/claude-skills/automode-edit.svg" alt="Measured on agentmods" height="20"></a>
Per session 43 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 2,335 The whole file, excluding the scripts and references it only reads on demand.
Security scan B 1 finding. 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 5.1 $0.00043 $0.02335
Opus 5 $0.00022 $0.01167
Sonnet 5 $0.00009 $0.00467
Haiku 4.5 $0.00004 $0.00233

Measured 6d ago against content hash 5795fcbde403, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-06, from the pricing page.

Security

Grade B, and why

automode-edit scanned grade B 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 6d 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.

Reads agent configuration directoriesmediumAgent snooping

.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.

`--write-shared`, never touch `~/.claude/settings.json`, unless the
skills/automode-config/commands/automode-edit.md · 230 lines

How it starts

The opening of the file, as written. The whole thing — 230 lines — stays where its author put it; the contents beside it link to each section on GitHub.

/automode-edit <query>

Translate a natural-language request into a structured modification of .claude/settings.local.jsonautoMode, then drive the existing six-phase apply pipeline. The agent is the interpreter; the deterministic guards (schema validation, mistaken-pattern detection, critique gate, hash gate, atomic write under flock) still apply.

<query> examples:

  • add a hard_deny rule that forbids pushing to main
  • allow deploys to the staging namespace
  • the github org is acme-corp, all repos under it are internal
  • drop the soft_deny entry about npm install
  • take ownership of the allow section (remove $defaults)
  • restore $defaults in soft_deny
  • rename the rule about migrations to mention the new ./scripts/migrate.sh path

Hard rules (do not violate)

  1. Prose, not patterns. Rules are natural-language sentences. If the user dictates Bash(...), Read(...), WebFetch(...), etc., refuse and explain that those belong in permissions, not in autoMode.
  2. Four sections only. environment, allow, soft_deny, hard_deny. Reject any other section. Migrate legacy denysoft_deny, drop ask with a warning.
  3. Never silently drop "$defaults". If the user's request implies removing the sentinel from a section, surface the consequence (which curated baseline they lose) and require an explicit confirm.
  4. Write target is .claude/settings.local.json only. Never --write-shared, never touch ~/.claude/settings.json, unless the user explicitly asks for it inside <query> and confirms again after the diff.
  5. No bypass of the pipeline. Always go through apply_automode.py --dry-run → user confirm → --approved-canonical-hash. Never write the JSON file directly.
  6. Read before edit. Re-read .claude/settings.local.json right before building the proposal; do not rely on cached content from earlier in the session.

Step 1 — load current state

Resolve paths and read state with the project's own scripts:

Read the full file on GitHub · 230 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. 6d ago First seen · 230 lines · 43 tokens per session scan B 5795fcbde403

Subscribe to this mod's changes

automode-edit is a command published in the GitHub repository obeone/claude-skills (3 stars, last pushed yesterday), licensed MIT. It adds 43 tokens to every session and 2,335 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.