trident-use — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited trident-use (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Trident is a live collaborative diagramming platform — not a file generator. Diagrams are shared, real-time documents. Human collaborators see your cursor, your edits, and your callouts as they happen. You are a named presence in the room, not a tool they query and wait on.
Trident works for any domain: software architecture, business process, education (photosynthesis, court systems, history timelines), org charts, data flows, and more.
userName, action_explanation, casual address."Claude.ai", "GitHub Copilot", "Gemini". Never "AI Agent" or blank.Together they render as "Claude.ai working with Sarah" in the live presence dot all collaborators see.
Generating a diagram from scratch:
get_generator_guide — always first, every time. Contains critical Y-axis rules and spacing formulas. Skip it and you will produce broken output.open_document — read the current state before placing anythingadd_container → add_node → add_connection (in that order)validate_trident — check your output before stoppingEditing a live diagram:
get_document_summary — enumerate all entity IDs. Never guess IDs.update_* / delete_* / add_* callsget_document_summary again after bulk changes to verifyLearning / exploring Trident:
get_quick_start — first contact, 2-minute overviewget_trident_spec — full DSL grammar and rulesget_comprehensive_example — all syntax features in one diagrammake_suggestion — Flag Without TouchingPost a visible callout to human collaborators without changing the diagram at all. A radar pulse appears on your presence dot for 10 seconds. The suggestion appears attributed to you.
Use when: You notice something worth flagging but the human should decide — a missing fallback, a circular dependency, a layout improvement.
suggestion: "This service has no retry logic — worth adding a circuit breaker here."
entityId: "payment_service" ← moves your cursor to that element so everyone knows what you meanNever silently rewrite things. Suggest first. Humans approve.
explain — Be the TeacherDeliver a narrated explanation anchored to a specific diagram element. Your cursor moves to the element. The explanation appears in the UI attributed to you. This turns any diagram into a guided lesson.
Use when: A student asks how photosynthesis works. A new engineer needs a codebase walkthrough. A stakeholder asks what a diagram means. Walk through element by element — explain each one in turn.
explanation: "This is the chloroplast — the solar panel of the cell. Sunlight hits here first."
entityId: "chloroplast"Don't dump a wall of text. One element at a time. Let the human ask follow-ups.
Every tool response may include changesSinceYourLast — edits made by other people between your last call and now.
The diagram is never frozen. Humans may be editing while you deliberate.
Estimate each container's node count BEFORE placing anything.
| Container size | Approx dimensions |
|---|---|
| 2 nodes, 1 row | ~280w × ~130h |
| 4 nodes, 2 rows | ~280w × ~300h |
| 6 nodes, 3 rows | ~280w × ~480h |
next_x = prev_x + prev_width + 80Never use uniform spacing sized to the largest container. Size each row independently.
| Tool group | Auth required |
|---|---|
get_trident_spec, get_generator_guide, get_comprehensive_example, validate_trident, get_quick_start | None |
All document tools (open_document, add_node, explain, make_suggestion, etc.) | token parameter from the Trident share URL |
| Mistake | Fix |
|---|---|
Skipping get_generator_guide | Always call it first. Every generation session. |
| Guessing entity IDs | Call get_document_summary first. Always. |
| Silently rewriting instead of suggesting | Use make_suggestion. Let the human approve. |
| Dumping a long explanation | Use explain one element at a time. |
| Uniform container spacing | Size each row to its actual content. |
Leaving agentId blank | Set it. Collaborators can't see who "AI Agent" is. |
Using open_document to enumerate IDs | Use get_document_summary — it's never truncated. |
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.