Build trust through clarity. Every word should advance understanding or deliver value. Cut the rest.
Respect the reader's time. Technical and strategic audiences are busy. Get to the point immediately.
Technical depth with human touch. We're writing for software engineering, security engineers and executives alike. Be precise without being pompous.
Show, don't tell. Examples and specifics beat abstractions every time.
- Conversational authority: Write like a trusted colleague explaining over coffee
- Humble expertise: We're all figuring this out together
- Urgently calm: The stakes are high but panic helps no one
- Community-first: "We" over "I", collaboration over competition
- Active voice dominates: "Agents expose events" not "Events are exposed by agents"
- Present tense when possible: "AOS provides" not "AOS will provide"
- Technical terms get defined once, then used freely
- Average 15-20 words per sentence
- Vary sentence length for rhythm
- Start with strong verbs
- End with impact
- Hook immediately: Lead with impact or bold assertion
- Context quickly: One sentence max for background
- Meat directly: Technical details with business translation
- Path clearly: Actionable next steps
- 2-4 sentences max: White space is your friend
- One idea per paragraph: Don't make readers multitask
- Topic sentences lead: State the point, then support it
- Transitions connect: "While X is true, Y presents challenges"
- Action-oriented: "Configure Guardian Agents" not "Guardian Agent Configuration"
- Specific over vague: "Trace Agent Decisions with OpenTelemetry" not "Observability Options"
- Value explicit: Include the "why" when possible
- Analogy first: "Guardian Agents are like security cameras for AI"
- Progressive disclosure: Simple → Complex → Implementation
- Business impact clear: Always answer "So what?"
- Examples concrete: Real scenarios beat abstract descriptions
# Bad: Wall of unexplained config
agent:
observability:
enabled: true
level: debug
# Good: Purposeful with context
# Enable full observability for development
agent:
observability:
enabled: true
level: debug # Captures all agent decisions- Impact leads: Business consequence before technical detail
- Threat specific: "Attackers can exfiltrate data via tool calls" not "Security vulnerabilities exist"
- Mitigation actionable: Exact steps, not general advice
- Timeline transparent: When discussing roadmap
- Throat-clearing: "In this document, we will explore..."
- Hedging: "It might be possible that perhaps..."
- Redundancy: "completely eliminate", "totally prevent"
- Academic flourishes: "It is important to note that..."
- Meta-commentary: "As mentioned above..."
- Adjectives and adverbs: Do they add precision?
- Passive constructions: Can you make it active?
- Long sentences: Can you split them?
- Technical jargon: Is there a simpler word?
- Lead with ASCII art or badges for credibility
- Installation → Configuration → Usage flow
- Platform-specific examples (Windows/Linux/macOS)
- Link liberally to related resources
- Personal anecdote or industry observation to open
- Technical meat with personality sprinkled throughout
- Forward-looking conclusion
- One clear call-to-action
- Precision over personality
- RFC-style numbering and structure
- Examples for every concept
- Interoperability considerations prominent
- Specific threat/issue (one sentence)
- Business impact (one sentence)
- Technical root cause (paragraph)
- Solution approach (bullets)
- What changed (one line)
- Why it matters (one line)
- How to use it (code example)
- What's next (one line)
- "trace" not "audit" (action-oriented)
- "folks" not "users" (human)
- "enable" not "allow" (empowering)
- "transparent" not "visible" (complete)
- "trustworthy" not "secure" (outcome-focused)
- Delivers, enables, provides, ensures
- Exposes, reveals, surfaces, demonstrates
- Prevents, blocks, denies, restricts
- Tracks, traces, monitors, observes
- Buzzwords without substance
- Military metaphors ("battle", "war room")
- Fear-mongering ("catastrophic", "devastating")
- Absolute claims ("always", "never") unless technically accurate
Before publishing any content:
- Impact clear in first paragraph?
- Technical accuracy verified?
- Business value articulated?
- Examples concrete and relevant?
- Next steps actionable?
- Tone conversational yet authoritative?
- Length reduced by 20%?
- Jargon defined or eliminated?
- Structure scannable with headers?
- Community perspective included?
"The implementation of comprehensive observability mechanisms within agentic systems represents a critical requirement for establishing trust in autonomous AI deployments across enterprise environments."
"Enterprises need to see what their AI agents are doing. Trust requires transparency."
"It should be noted that the configuration of Guardian Agents may potentially require administrative privileges depending on the deployment scenario under consideration."
"Guardian Agents need admin privileges in some deployments."
"In order to facilitate the comprehensive tracing of agent interactions, organizations must implement standardized telemetry collection."
"Track every agent action. Use standard telemetry. Here's how:"
Great writing disappears. Readers should focus on ideas, not prose.
Clarity beats cleverness. That brilliant metaphor? Cut it.
We're building together. This is a community effort. Write like it.
Remember: Good writing is rewriting. Great writing is deleting.