llm-privacy
Word add-in that checks for consistency across the document and can set a style by training on a style sample
Install / Use
npx skills add Gekko47/ToneForgeInstalls into whichever agent you are using.
.clinerules
Cline rules
Quality Score
Category
AI & Machine LearningSupported Platforms
Tags
Our assessment of llm-privacy
llm-privacy scores 56/100 on our quality scale, 906th of 947 AI & Machine Learning skills we index.
Its .clinerules is 3.3 KB long, well organised into 9 sections and no code examples: a solid amount of guidance for an agent.
It has no GitHub stars yet, so there is no community track record; judge it on its content.
Maintenance, license and trust
- We could not determine when the repository was last updated.
- Our last check on 2026-11-10 found the source still online.
- No license is declared. By default that means all rights are reserved: you can read it, but reusing or redistributing it is not clearly permitted. Ask the author before building on it commercially.
- Its trust signals score 68/100, with 3 cautions from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.
Safety scan
No issues foundOur scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.
AI review by kimi-k2.7-code on 2026-10-02. Automated pattern scan on 2026-10-01. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.
llm-privacy compared with similar skills
All 4 of these similar skills score higher than llm-privacy; compare them before choosing.
| Skill | Score | Stars | Updated | Format |
|---|---|---|---|---|
| llm-privacy (this skill)by Gekko47 | 56 | 0 | — | .clinerules |
| claude-memby thedotmack | 100 | 117.6k | today | CLAUDE.md |
| Understand-Anythingby Egonex-AI | 100 | 92.8k | 1d ago | CLAUDE.md |
| headroomby headroomlabs-ai | 100 | 79.8k | today | CLAUDE.md |
| CowAgentby zhayujie | 100 | 48.0k | today | CLAUDE.md |
Frequently asked questions
- How do I install llm-privacy?
- Run
npx skills add Gekko47/ToneForge. The install tabs above show the steps for each supported agent. - Which AI agents does llm-privacy work with?
- It is written for Cline, as a .clinerules file. Other agents that read the same format can often use it too.
- Is llm-privacy safe to use?
- Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It declares no license and scores 68/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
- Is llm-privacy still maintained?
- We could not determine when the repository was last updated.
Skill content
View source on GitHubpaths:
- "src/ai/**"
- "src/core/config/**"
- "src/core/domain/ProviderConnection.ts"
- "src/taskpane/components/OpenRouterConnectionSettings.tsx"
- "src/taskpane/components/ProviderPrivacySettingsSection.tsx"
- ".env"
- ".env.example"
- "scripts/dev-broker.mjs"
- "scripts/dev-gateway.mjs"
LLM Privacy and Credentials
Raw document text must never leave the add-in without explicit opt-in, and no API key is ever persisted by the add-in.
Rules
1. Prompt builders throw unless raw text is explicitly opted in
Every builder in src/ai/prompts/ takes an
includeRawText option and throws when it is not true. Never bypass this. A
new prompt builder gets the same gate. Validate every response with a Zod
schema; never trust raw model output.
2. The add-in never persists a credential
Not in Office.roamingSettings, not in localStorage, not in the bundle, not in
a URL, not in a log line.
ProviderConnection.tshas no field capable of holding a secret. A test reflects over the schema so a future field cannot quietly reintroduce one. If your change needs a credential, the architecture is wrong, not the test.- The browser-held
apiKeycredential mode was removed (ADR-0049). Do not reintroduce it. env.tsholds no secret field by design. Webpack compiles only an allowlisted, non-secret set. The gitignored.envis read only by the Node-side development broker.
3. OpenRouter's key is transient and gateway-held
OpenRouter is the one provider taking a user-held key. It lives in component
state only, is submitted once over the loopback same-origin and
nonce-protected channel, and is dropped when the request settles, success or
failure. Only the opaque connection reference is persisted, and the key is held
in the local development gateway's memory (ADR-0050). Keep it in
OpenRouterConnectionSettings.tsx, a separate component, so it cannot leak into
LlmSettingsDraft — which has no field that could hold one.
4. One consent governs one feature
semanticOptIn and consistencyReviewConsent are independent; the spot and
full-document consents are retired and kept only for migration. None implies
another, consistencyReviewConsent defaults to false, and migration sets it to
false rather than deriving it. A change letting one consent enable another
contradicts ADR-0052.
5. Redact secrets and content everywhere
redactSensitiveText in
src/shared/utils/redaction.ts is the
shared implementation. Credential-shaped, prompt-content, and document-content
fields are redacted from logs and recursive error context. Honor AbortSignal:
caller cancellation is non-retryable, an internal timeout is retryable.
6. Never commit secrets
.env is gitignored; .env.example is the committable template.
npm run secrets:scan and npm run secrets:verify-build are part of the
verification graph and will fail the commit.
Referenced resources
- docs/privacy-security.md — full posture
- docs/decision-log.md — ADR-0049, ADR-0050, ADR-0052
- .cline/skills/toneforge-llm/SKILL.md
Related Skills
claude-mem
117.6kPersistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More
Understand-Anything
92.8kGraphs that teach > graphs that impress. Turn any code into an interactive knowledge graph you can explore, search, and ask questions about. Works with Claude Code, Codex, Cursor, Copilot, Gemini CLI, and more.
headroom
79.8kCompress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
CowAgent
48.0kOpen-source personal AI assistant & Agent Harness. Plans tasks, runs tools and skills, self-evolves with memory and knowledge. Multi-agent, multi-model, multi-channel. Lightweight, extensible, one-line install.
Trust signals
From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.
