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.
npx agentmods add commands/xbluesky/cortexes/querygit clone --depth 1 https://github.com/XBlueSky/cortexesWrote 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/xbluesky/cortexes/query)<a href="https://agentmods.dev/commands/xbluesky/cortexes/query"><img src="https://agentmods.dev/badge/commands/xbluesky/cortexes/query.svg" alt="Measured on agentmods" 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 | $0.00018 | $0.00651 |
| Opus 5 | $0.00009 | $0.00326 |
| Sonnet 5 | $0.00004 | $0.00130 |
| Haiku 4.5 | $0.00002 | $0.00065 |
Grade A, and why
query 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 today.
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.
What it actually says
Invoke the cortexes:cortex-query skill and follow it for the actual search.
Command frontmatter cannot load a skill for you, so invoke it explicitly with
its fully qualified name via the Skill tool.
Argument handling:
- With arguments → treat
$ARGUMENTSas the search query verbatim. Do not reword or narrow it; the user's phrasing is the query. - No arguments → ask the user what to search for, then run the search with their answer. Do not guess a query from the conversation, and do not search the whole vault "to see what's there".
This command is user-invoked only (disable-model-invocation: true), so
running it is the user's explicit request — using-cortex signal 1. Search
even if the session earlier picked "直接開始工作" at SessionStart.
Follow the skill's layered strategy: cortex-vec search first (repo-scoped by
default inside a git repo, overridable with "search all"), exact-match search
as the supplement, and Raw/ only on request. Present results in the skill's
response format, and read a full page only when the user picks one.
This command pre-approves only what the search itself runs — cortex-vec search, grep, and the two read-only git calls used to detect the repo —
rather than shell access in general. Each entry is scoped to the subcommand:
git remote get-url, not git remote, which would also cover add,
remove, rename and set-url — write operations a read-only search has no
business pre-approving. cortex-vec status is deliberately absent; the search
flow never calls it. grep is on the list because the vault usually sits
outside the session's working directory, where the Grep tool cannot reach.
Scoped Bash does not escape the workspace boundary the way a blanket Bash
approval did. cortex-vec search is unaffected (it takes no path argument),
so the normal path still works; only the grep fallback needs the vault to be
inside the working directory or added with /add-dir <vault_path>. If that
bites, say so explicitly rather than reporting an empty result.
If cortex-vec is not installed, say so and fall back to grep rather than
failing. Semantic scoring needs OPENAI_API_KEY; without it cortex-vec
degrades to its local BM25 index on its own, which is still a real search —
report the degraded mode, don't call retrieval unavailable.
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.
- today First seen · 56 lines · 18 tokens per session scan A dfc273781483
query is a command published in the GitHub repository XBlueSky/cortexes (20 stars, last pushed 3d ago), licensed Apache-2.0. It adds 18 tokens to every session and 651 once invoked, about $0.0001 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 2026-09-04.
Other commands, from other repositories
search-code
Hybrid code search — semantic + keyword + BM25.
index
Visual overview of Beacon index — chunks, coverage, provider. Pass --files to list indexed files.
config
View and modify Beacon configuration — providers, thresholds, indexing rules.
blacklist
Manage Beacon indexing blacklist — prevent indexing of dangerous directories.
whitelist
Manage Beacon indexing whitelist — allow indexing in specific directories that would otherwise be blacklisted.
run-indexer
Manually trigger Beacon indexing (useful when auto-index is off).