ソース情報
- リポジトリ
- ci/persops
- ソースの最終更新活動
- 2026年7月31日 23:31
- 検出された SKILL.md の言語
- 英語
- スター
- 0
- フォーク
- 0
インストール方法
デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。
ソースファイルを確認
インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。
メニュー
デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。
インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。
SOC 職業分類に基づく
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
直接コマンドでは確認用 Prompt が省略されます。実行前にソースを確認してください。
npx skills add https://github.com/ci/persops --skill one-passwordコマンドは1行のまま表示されます。コピー前に横へスクロールして全体を確認してください。
ローカルで確認しますか?SkillsMP が現在取得できるファイルをダウンロードできます。
SKILL.md を表示中
Deploys persops configurations to aglaea and amalthea. Use when asked to build, apply, switch, or deploy persops locally or through an Amp runner.
Add or update agent skills in ~/p/persops, including vendoring upstream skills into modules/ai/skills, choosing the right propagation profile (`all`, `coding`, `claw`, `codex`), updating metadata, and applying local+remote Nix switches. Use when the user asks to add a skill, install a skill, update a skill, make a coding skill, make a Codex-only skill, or make a claw/openclaw/clawdbot-only skill in persops.
Install and configure the anti-slop Oxlint plugin in a local TypeScript or JavaScript repository. Use whenever a user asks to add anti-slop lint rules, copy the anti-slop plugin, configure opinionated Oxlint rules, or migrate an existing local anti-slop setup.
| name | one-password |
| description | 1Password/op: service-account first, targeted secret read/store/inject, tmux. |
| metadata | {"clawdbot":{"emoji":"🔐","requires":{"bins":["op","tmux"]},"install":[{"id":"brew","kind":"brew","formula":"1password-cli","bins":["op"],"label":"Install 1Password CLI (brew)"}]}} |
Follow the official CLI get-started steps. Don't guess install commands.
references/get-started.md (install + app integration + sign-in flow)references/cli-examples.md (real op examples, including safe item create/edit patterns)op --version.op signin once inside the same session. The 1Password prompt is the user approval boundary; ask in chat only if the prompt cannot be surfaced or completed.op whoami.--account or OP_ACCOUNT.tmux send-keys; do not start a second session just to retry.my.1password.com.my.1password.eu / Titan unless explicitly asked.--account my.1password.com on every op command when storing or reading secrets. Do not rely on ambient account selection.op account list is metadata-only, but still must run inside tmux. Use it to confirm account names when routing is unclear.op signin --account my.1password.com can return status 0 with no useful output and still not make a later shell signed in. Prefer doing sign-in, create/edit/get, and verification in the same tmux shell.~/.profile exports OP_SERVICE_ACCOUNT_TOKEN, use it for scoped service-account commands.OP_SERVICE_ACCOUNT_TOKEN="$OP_SERVICE_ACCOUNT_TOKEN" op item get "<known item>" --vault "<known vault>" ....op reads require an explicit vault query; omitting --vault can fail even when the token is valid.op command, including service-account reads, still runs inside one named tmux session.The shell tool uses a fresh TTY per command. Run op inside one dedicated tmux session and keep using that same session until the whole secret task is done. Service-account commands still run here, but must not trigger app prompts.
Example:
SOCKET_DIR="${CLAWDBOT_TMUX_SOCKET_DIR:-${TMPDIR:-/tmp}/clawdbot-tmux-sockets}"
mkdir -p "$SOCKET_DIR"
SOCKET="$SOCKET_DIR/clawdbot-op.sock"
SESSION="op-work"
tmux -S "$SOCKET" has-session -t "$SESSION" 2>/dev/null ||
tmux -S "$SOCKET" new -d -s "$SESSION" -n shell
tmux -S "$SOCKET" send-keys -t "$SESSION:" -- "op signin --account my.1password.com" Enter
tmux -S "$SOCKET" send-keys -t "$SESSION:" -- "op whoami" Enter
tmux -S "$SOCKET" capture-pane -p -J -t "$SESSION:" -S -200
Do not create a new tmux session after a quoting, item-name, or command failure. Send a corrected command into the existing session.
Target the session as $SESSION: instead of assuming window 0; older sessions may have window indexes starting at 1.
$npm; it documents npm-specific auth flow and package reservation helper.op, targeted reads, one persistent session, no broad enumeration, no secret output.Use the persistent tmux session. Write the exact secret task to a temp script, then send that script into op-work; do not create a second tmux session for retries.
SOCKET_DIR="${CLAWDBOT_TMUX_SOCKET_DIR:-${TMPDIR:-/tmp}/clawdbot-tmux-sockets}"
SOCKET="$SOCKET_DIR/clawdbot-op.sock"
SESSION="op-work"
tmux -S "$SOCKET" has-session -t "$SESSION" 2>/dev/null ||
tmux -S "$SOCKET" new -d -s "$SESSION" -n shell
cat > /tmp/op-store-secret.sh <<'SCRIPT'
#!/usr/bin/env bash
set -euo pipefail
set +x
ACCOUNT="my.1password.com"
ITEM_TITLE="Service API Tokens"
FIELD_NAME="api_token"
EXPECTED_PREFIX=""
NOTES="Created via tmux-safe op workflow"
TOKEN="$(pbpaste)"
if [ -n "$EXPECTED_PREFIX" ]; then
case "$TOKEN" in "$EXPECTED_PREFIX"*) ;; *) echo "clipboard value does not match expected prefix" >&2; exit 2;; esac
fi
op item create --account "$ACCOUNT" --category "API Credential" --title "$ITEM_TITLE" "$FIELD_NAME[password]=$TOKEN" "notesPlain=$NOTES" >/dev/null
op item get --account --fields >/dev/null
SCRIPT
700 /tmp/op-store-secret.sh
tmux -S send-keys -t -- C-m
The op category string is human-readable and case-sensitive in this CLI build; use "API Credential", not api_credential.
For a known item, verify the field shape before using it live: length, expected prefix, newline count, never value. op --field NAME and --fields label=NAME can return the wrong concealed field when an item has duplicate/legacy credential fields. If shape is wrong, read the known item as JSON and extract the exact label.
cat > /tmp/op-read-field.sh <<'SCRIPT'
#!/usr/bin/env bash
set -euo pipefail
set +x
ITEM_TITLE="Known API Credential Item"
FIELD_LABEL="api_token"
VAULT="Known Vault"
value="$(
OP_SERVICE_ACCOUNT_TOKEN="$OP_SERVICE_ACCOUNT_TOKEN" \
op item get "$ITEM_TITLE" --vault "$VAULT" --format json |
FIELD_LABEL="$FIELD_LABEL" node -e 'let s=""; process.stdin.on("data",d=>s+=d); process.stdin.on("end",()=>{const item=JSON.parse(s); const f=(item.fields||[]).find(x=>x.label===process.env.FIELD_LABEL); if(!f?.value) process.exit(2); process.stdout.write(f.value);})'
)"
echo "field_len:${#value}"
case "$value" in sk-*) echo "field_prefix:sk" ;; *) echo "field_prefix:other" ;; esac
echo "field_has_newline:$(printf %s "$value" | wc -l | tr -d ' ')"
SCRIPT
chmod 700 /tmp/op-read-field.sh
tmux -S "$SOCKET" send-keys -t "$SESSION:" -- "bash /tmp/op-read-field.sh; rm -f /tmp/op-read-field.sh" C-m
Keep JSON extraction scoped to the known item and vault. Do not enumerate vaults/items to discover candidates.
Only use this when the user explicitly asks to search, gives a screenshot/listing, or the exact title guess failed and the user asks for regex/fuzzy lookup. Stay vault-scoped and metadata-only; print candidate titles/ids/categories/vault names, never fields or values. Prefer exact visible strings from screenshots first: vault name, item title, and field label.
cat > /tmp/op-find-item.sh <<'SCRIPT'
#!/usr/bin/env bash
set -euo pipefail
set +x
VAULT="Known Vault"
QUERY="minimax"
OP_SERVICE_ACCOUNT_TOKEN="$OP_SERVICE_ACCOUNT_TOKEN" \
op item list --vault "$VAULT" --format json |
QUERY="$QUERY" VAULT="$VAULT" node -e '
let s=""; process.stdin.on("data",d=>s+=d); process.stdin.on("end",()=>{
const q=process.env.QUERY.toLowerCase();
const vault=process.env.VAULT;
const items=JSON.parse(s).filter(x => [
x.title, x.id, x.category, ...(x.tags || [])
].filter(Boolean).join("\n").toLowerCase().includes(q));
for (const item of items.slice(0, 10)) {
console.log(`title:${item.title} id:${item.id} category:${item.category || ""} vault:${vault}`);
}
console.log(`matches:${items.length}`);
})'
SCRIPT
chmod 700 /tmp/op-find-item.sh
tmux -S "$SOCKET" send-keys -t "$SESSION:" -- "bash /tmp/op-find-item.sh; rm -f /tmp/op-find-item.sh" C-m
After choosing a candidate, switch back to exact item/field JSON extraction and shape-only validation. An exact known personal item may use desktop fallback automatically; do not broadly enumerate personal vaults unless the user asked to search.
Keep the whole pipeline inside the same tmux session. Inspect status and output length, never secret values.
cat > /tmp/op-debug.sh <<'SCRIPT'
#!/usr/bin/env bash
set -euo pipefail
set +x
SIGNIN_OUTPUT="$(op signin --account my.1password.com 2>&1 || true)"
echo "signin output bytes: ${#SIGNIN_OUTPUT}"
op account list 2>&1 | sed -E "s/(xox[baprs]-)[A-Za-z0-9-]+/\\1REDACTED/g; s/(xapp-)[A-Za-z0-9-]+/\\1REDACTED/g"
SCRIPT
chmod 700 /tmp/op-debug.sh
tmux -S "$SOCKET" send-keys -t "$SESSION" -- "bash /tmp/op-debug.sh; rm -f /tmp/op-debug.sh" C-m
op run / op inject over writing secrets to disk.op account add.op signin inside tmux and authorize in the app.op outside tmux; stop and ask if tmux is unavailable.