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 skills/johnnywuj81/tokenknows/tokenknows-knowledgenpx skills add johnnywuj81/tokenknows --skill tokenknows-knowledgegit clone --depth 1 https://github.com/johnnywuj81/tokenknowsWrote 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/skills/johnnywuj81/tokenknows/tokenknows-knowledge)<a href="https://agentmods.dev/skills/johnnywuj81/tokenknows/tokenknows-knowledge"><img src="https://agentmods.dev/badge/skills/johnnywuj81/tokenknows/tokenknows-knowledge.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 | $0.00084 | $0.00521 |
| Opus 5 | $0.00042 | $0.00260 |
| Sonnet 5 | $0.00017 | $0.00104 |
| Haiku 4.5 | $0.00008 | $0.00052 |
Grade A, and why
tokenknows-knowledge 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 4d 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.
What it actually says
TokenKnows 知识检索
查询跨会话沉淀的项目知识资产。工具来自 tokenknows MCP server。
何时用
- 用户想回溯历史决策:"我们之前为什么选 ClickHouse"、"有没有相关 ADR"
- 用户想查实体:"X 模块谁改过"、"搜一下 Y 这个概念"
- 用户想看知识库里有什么:"列一下已生成的文档"
关键步骤
- 列资产 (use
mcp__tokenknows__list_assets): 按 type / 时间窗筛选项目里已蒸馏的文档 (周报 / ADR / KG / ...)。 - 读详情 (use
mcp__tokenknows__get_asset+mcp__tokenknows__get_asset_chapters): 拿到某份文档的章节正文 + 证据链。 - 跨文档实体搜索 (use
mcp__tokenknows__search_entity): 传query(+ 可选entity_type: person / concept / artifact / event), 在知识图谱里找实体 + 它出现在哪些文档。
好例子 / 坏例子
- 坏:用户问"我们之前怎么决定数据库的"→ 凭印象答。应
search_entity("数据库选型")或list_assets(type="adr")找到真实 ADR 再答,带出处。 - 好:
search_entity(query="ClickHouse")→ 返回相关实体 + 所在 ADR/KG → 给用户带链接的答案。
注意
- 只读操作,不改知识库。
- 依赖本地后端 (
http://127.0.0.1:8001) 在跑。后端不通 / 401 / 404 时,工具错误信息已含修复指引,原样转述给用户即可。
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.
- 4d ago First seen · 36 lines · 84 tokens per session scan A 07a36363ade4
tokenknows-knowledge is a skill published in the GitHub repository johnnywuj81/tokenknows (4 stars, last pushed yesterday), licensed MIT. It adds 84 tokens to every session and 521 once invoked, about $0.0004 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 2026-08-31.
Other skills, from other repositories
agentline
Make phone calls, view received SMS, provision numbers, manage agents, and track billing through the AgentLine telephony API (REST or MCP). Use when the user asks to call someone, check transcripts, view text messages, manage phone agents, buy numbers, or check account balance. For MCP-native workflows, the server at…
excalidraw-ai
Create professional Excalidraw diagrams by generating JSON directly. This skill provides the Excalidraw JSON schema reference and professional icon libraries for AI agents to autonomously create diagrams without templates.
error-handling
Python error handling patterns for FastAPI, Pydantic, and asyncio. Follows "Let it crash" philosophy - raise exceptions, catch at boundaries. Covers HTTPException, global exception handlers, validation errors, background task failures. Use when: (1) Designing API error responses, (2) Handling RequestValidationError…
logfire
Structured observability with Pydantic Logfire and OpenTelemetry. Use when: (1) Adding traces/logs to Python APIs, (2) Instrumenting FastAPI, HTTPX, SQLAlchemy, or LLMs, (3) Setting up service metadata, (4) Configuring sampling or scrubbing sensitive data, (5) Testing observability code.
linting
Python linting with Ruff - an extremely fast linter written in Rust. Use when: (1) Standardizing code quality, (2) Fixing style warnings, (3) Enforcing rules in CI, (4) Replacing flake8/isort/pyupgrade/autoflake, (5) Configuring lint rules and suppressions.
commit-message
Analyze git changes and generate conventional commit messages. Supports batch commits for multiple unrelated changes. Use when: (1) Creating git commits, (2) Reviewing staged changes, (3) Splitting large changesets into logical commits.