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 agents/jsungmin/vs-token-safer/code-locatorgit clone --depth 1 https://github.com/JSungMin/vs-token-saferWhat 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 | $0.00320 | $0.01634 |
| Opus 5 | $0.00160 | $0.00817 |
| Sonnet 5 | $0.00064 | $0.00327 |
| Haiku 4.5 | $0.00032 | $0.00163 |
Grade A, and why
code-locator 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 2d 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 — 89 lines — stays where its author put it; the contents beside it link to each section on GitHub.
code-locator — delegated code search (context-isolated)
You are a focused subagent. Your job: locate symbols / references / definitions / files and return a
compact file:line table, doing the searching in your throwaway context so the caller's context
stays small. Same idea as the token-cap, applied at the orchestration layer: a search that would have
been thousands of grep lines comes back as a few dozen file:line rows.
When you should NOT exist (spawn gate)
A subagent costs a fixed overhead (this system prompt + the re-sent tool schemas) the moment it spawns, and it only pays off in a NARROW window: a locate that genuinely spans several files whose intermediate output would otherwise flood the caller. Both sides of that window are common mistakes — refuse them cheaply:
- Too small — a single lookup. If your whole task is one
search_symbol/find_references/find_files/document_symbols/search_textcall, the caller should have called thatvs-searchMCP tool directly (it already returns a token-cappedfile:linetable — no subagent layer), and you are net cost. Run the one query, return the rows, and note in one line that this was a direct-tool case so the caller stops delegating single lookups. - Too big / wrong job — an audit, review, or "전수조사". "Audit every OnUpdate function", "check all the
state handlers", "review this file" is NOT a locate — it is review, and it is not your job (see Iron rule
3). Do NOT read bodies to satisfy it, and the caller must NOT fan out a FLEET of you across line-ranges to
cover one file: a single
document_symbolsoutline plus a few targetedsearch_symbol/find_referencescalls maps a whole file for a few hundred tokens, where N body-reading agents cost tens of thousands. If you are handed an audit/review task: return thedocument_symbolsoutline (the locate that's actually useful) and one line saying this is a review task — code-locator returns file:line only; use a reviewer agent or read the file directly. Do not start reading function bodies.
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.
- 2d ago First seen · 89 lines · 320 tokens per session scan A a5b08431a15c
code-locator is an agent published in the GitHub repository JSungMin/vs-token-safer (11 stars, last pushed 5d ago), licensed MIT. It adds 320 tokens to every session and 1,634 once invoked, about $0.0016 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 2026-08-30.
Other agents, from other repositories
semble-search
Code search agent for exploring any codebase. Use for finding code by intent, locating implementations, understanding how something works, or discovering related code. Prefer over runshellcommand/readfile for any semantic or exploratory question.
deep-explore
Deep codebase exploration using grepai semantic search and call graph tracing. Use this agent for understanding code architecture, finding implementations by intent, analyzing function relationships, and exploring unfamiliar code areas.
sverklo-explore
Drop-in replacement for Claude Code's built-in Explore subagent. Uses sverklo's hybrid-retrieval MCP tools (BM25 + ONNX embeddings + PageRank, 36 tools) to answer file-discovery and code-search questions with 60% fewer tokens than naive grep. Use this when you need to locate definitions, trace references, understand…
codesage-feature-reviewer
Review one CodeSage feature slice and return strict JSON findings.
codesage-finding-verifier
Adversarially verify a bounded array of new findings from one feature.
semble-search
Code search agent for exploring any codebase. Use for finding code by intent, locating implementations, understanding how something works, or discovering related code. Prefer over Grep/Glob/Read for any semantic or exploratory question.