personalize-auto-mode
Rewrite an existing Claude Code autoMode block against this machine's facts: re-point the names, organizations, SSO profiles, repositories and paths in the environment paragraphs from ~/.config/auto-mode-tools/facts.yml, then write the live settings.json after a diff, an approval, a backup and a validation pass.
- License 许可
- GPL-3.0
- 最近更新
- 2026-11-14
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- personalize-auto-mode
- description
- Rewrite an existing Claude Code autoMode block against this machine's facts: re-point the names, organizations, SSO profiles, repositories and paths in the environment paragraphs from ~/.config/auto-mode-tools/facts.yml, then write the live settings.json after a diff, an approval, a backup and a validation pass.
03 SKILL.md 正文
You are running the /auto-mode-tools:personalize-auto-mode skill.
Take the autoMode block already in the user's Claude Code
settings.json and re-point it at this machine's reality: the names,
organizations, SSO profiles, repositories and paths that appear inside
the environment paragraphs.
This is seeded-only, not fully derived. The block being rewritten
is the seed and carries the reasoning; the facts file supplies what is
specific to this machine. After this skill runs, settings.json is the
living artifact and the facts file is expected to go stale. Re-running
the skill is a start-over, not an increment.
The facts file
$XDG_CONFIG_HOME/auto-mode-tools/facts.yml, defaulting to
~/.config/auto-mode-tools/facts.yml.
Read it and check schema-version. It must be 1. On any other
value, abort, naming the file and the version expected:
<facts-path> has schema-version <found>; this skill reads
schema-version 1. Aborting rather than reading it as if it matched.
If the file does not exist, stop and tell the human where to write it, showing the schema below. Do not invent facts and do not read them out of the block you are rewriting.
The schema, with placeholder values — no machine's real facts ship in
this plugin. The plugin's README.md carries this same block
verbatim, so a schema change edits both copies:
schema-version: 1
identity:
display-name: "Ada Lovelace"
github-login: alovelace
email: ada@example.org
bots:
- display-name: "Claude for Ada"
github-login: claude-for-alovelace
email: ada+claude@example.org
organizations: [ExampleOrg, OtherOrg]
sso-profiles: [example-dev, example-prod]
public-repositories:
- ExampleOrg/public-thing
scratch-globs: ["~/scratch/**"]
worktree-globs: ["**/.claude/worktrees/**"]
notes: |
Free prose for what the keys above cannot enumerate.
The split between the structured keys and notes: is deliberate. The
structured keys are the enumerable facts: they appear verbatim
inside the environment paragraphs and must be reproduced exactly, since
free prose invites dropping a list member or paraphrasing a profile
name with nothing to catch it. notes: is for what the keys cannot
enumerate.
identity and each bots entry carry the same three key names, so
read one shape twice rather than two shapes. A bots entry is one
record because a rule matching a bot generally needs its login and its
email to agree.
Procedure
Write the same file /auto-mode-tools:tune-auto-mode writes, so carry
the same protections, in this order:
- Read the live
~/.claude/settings.json. If it has noautoModeblock, stop and say so — this skill rewrites a block, it does not author one. - Produce the rewritten block in a scratch
CLAUDE_CONFIG_DIRcopy of that file, never in the live file. - Show the human the diff and get an explicit approval.
- Take a timestamped backup of the live file, and tell the human its path.
- Verify the candidate passes
CLAUDE_CONFIG_DIR=<scratch> claude auto-mode configwithout error. - Only then write the live file.
If either the approval or the validation does not come, leave the live file untouched and say where the scratch copy is.
What a rewrite is
The environment entries are dense paragraphs, not templates, so
personalizing one is a rewrite rather than a token substitution.
The structured keys give you exact lists to place; notes: and the
existing prose give you the rest to reason from. A paragraph that
mentions an organization the facts file does not list is a paragraph to
re-reason about, not a paragraph to leave standing with a stale name in
it.
Keep the block to environment, allow, soft_deny and hard_deny.
Independence
This skill does not depend on /auto-mode-tools:tune-auto-mode and
does not read the ledger. It writes nothing to the state directory.
It must work on an autoMode block that has never been tuned. There is
no ordering contract between the two skills.
Writing nothing there includes the tuner's
~/.local/state/auto-mode-tools/provenance.yml: a rewrite by this
skill leaves that file naming the revision the tuner last wrote, which
this rewrite has changed. That is deliberate — bumping the revision
here would claim a critique round this skill never ran, and reaching
into the state directory to repair the mismatch would end the
independence this section states. Say so if the human asks what
revision their block is at after a rewrite.
04 同仓库更多 Skills
mask-inappropriate-language
Replace inappropriate language (profanity, slurs, strong insults) in provided text or a file with asterisks, character-for-character, preserving everything else.
theorem-agents-interface
The interface between the sdlc reviewer and the theorem agents it spawns — what each double-dash brief parameter means, and what each consequence class means. Preloaded into every theorem agent (sdlc:theorem-generator, the disprover, and the verifier) via its skills frontmatter, and read by name by the sdlc:theorem-based-pr-reviewer agent for what its briefs' parameters mean; not invoked from the user's slash menu.
sdlc-config-user
Interactively create or merge-update the repo user sdlc config, `<repo-root>/.sdlc/user-config.yml` (tier 3 — this user, this repo, never committed), and add it to the repo's tracked `.gitignore`.
sdlc-config-repo
Interactively create or merge-update the shared repo sdlc config, `<repo-root>/.sdlc/repo-config.yml` (tier 2 — tracked, every user of this repo).
sdlc-config-global
Interactively create or merge-update the global user sdlc config, `$XDG_CONFIG_HOME/sdlc/user-config.yml` (tier 1 — this user, every repo on this machine).
pr-read-cli-interface
The contract for sdlc's PR-read CLIs — sdlc-pr-round, sdlc-pr-adjustments, sdlc-fixer-brief and sdlc-records-chain — their arguments, exit codes and output grammar. Preloaded into sdlc:theorem-based-pr-reviewer, sdlc:issue-fixer, sdlc:docs-writer and sdlc:pr-finalizer via their skills frontmatter; not invoked from the user's slash menu.