Professional Communication Agent Skill

Professional Communication Skill

Overview

This skill transforms dense technical communication into clear, structured business formats using proposition extraction (identify all facts and relationships) and deterministic templates (apply consistent structure). It extracts every detail without loss, categorizes by business relevance, applies a standard template with professional tone, and verifies completeness before delivery.

Core principle: Transformation ≠ creation. Only restructure existing input; always extract from existing input and restructure it for executive clarity with preserved technical accuracy.

Reference Loading Table

| Signal | Load These Files | Why |

|---|---|---|

| transforming raw engineer notes: stream-of-consciousness debugging, blockers, crisis updates | examples.md | Loads detailed guidance from examples.md. |

| drafting from section templates; phrase transformations | templates.md | Loads detailed guidance from templates.md. |

Instructions

Phase 1: PARSE

Goal: Extract every proposition from the input before structuring anything. This prevents information loss and ensures technical accuracy is preserved.

Step 1: Classify input type

Identify the communication type (this determines categorization strategy in Phase 2):

• Technical update (progress report with embedded facts)
• Debugging narrative (stream-of-consciousness problem-solving)
• Status report (project state with blockers/dependencies)
• Dependency discussion (constraints buried in defensive language)

Step 2: Extract all propositions

Parse each sentence systematically. Extract all propositions before summarizing — summarizing skips propositions and loses facts:

• Facts: All distinct statements of truth
• Implications: Cause-effect relationships
• Temporal markers: Past/present/future actions
• System references: All mentioned components
• Blockers: Hidden dependencies and constraints
• Emotional context: Frustration/satisfaction/urgency indicators (needed to transform defensive language)

Step 3: Document implicit context

Surface assumptions the author takes for granted but the audience needs stated. Non-technical audiences cannot act without this:

• Technical acronyms or project names the audience may not know
• Timeline context (when things happened relative to milestones)
• Organizational context (team relationships, reporting structures)

Step 4: Count and validate propositions

## Parsing Result
Input type: [technical update | debugging narrative | status report | dependency discussion]
Proposition count: [N distinct facts/claims]
Emotional markers: [frustration | satisfaction | urgency | neutral]

Extracted Propositions:
1. [Fact/claim 1]
2. [Fact/claim 2]
... (ALL propositions - NO information loss)

Implicit Context:
- [Assumption 1]
- [Assumption 2]

Gate: ALL propositions extracted with zero information loss. Proceed only when gate passes.

Phase 2: STRUCTURE

Goal: Categorize and prioritize all extracted propositions by business relevance. This prevents unsolicited sections and keeps output focused on what matters most.

Step 1: Categorize propositions

Organize by type (categorization determines template section placement):

Status:   [items with current state]
Actions:  [completed, in-progress, planned]
Impacts:  [business and technical consequences]
Blockers: [dependencies, constraints]
Next:     [required actions]

Step 2: Priority order

Rank by impact to executive decision-making, not completeness:

• Business Impact (revenue, customer, strategic)
• Technical Functionality (core operation)
• Project Timeline (schedule implications)
• Resource Requirements (personnel, infrastructure)
• Risk Management (potential issues)

Only the highest-priority categories go into the output. Lower-priority items are preserved in Technical Details but not emphasized.

Step 3: Identify information gaps

Flag any propositions that need clarification before transformation. Ask for specifics only when severity classification is ambiguous:

• Ambiguous severity (could be GREEN or YELLOW — default to YELLOW if unclear)
• Missing ownership for action items (block on clarity, ask for clarity)
• Undefined technical terms critical to business impact (ask for definition)

Gate: All propositions categorized and prioritized. Proceed only when gate passes.

Phase 3: TRANSFORM

Goal: Apply standard template with professional tone. This ensures consistent, executive-ready formatting without speculative sections.

Step 1: Apply standard template

Include only the sections in the standard template (Risk Assessment, Historical Context, Mitigation Strategies). Use ONLY this structure:

**STATUS**: [GREEN|YELLOW|RED]
**KEY POINT**: [Single most important business takeaway]

**Summary**:
- [Primary accomplishment/issue]: [Business impact]
- [Current focus/blocker]: [Expected outcome/resolution need]
- [Secondary consideration]: [Implications]

**Technical Details**:
[2-3 sentences maximum preserving technical accuracy]

**Next Steps**:
1. [Specific action with timeline if available]
2. [Secondary action with ownership implications]
3. [Follow-up considerations]

Step 2: Tone adjustment

The transformation rules are deterministic (apply all):

• Strip hedging language: "I think we might need to..." → "Deploy X to address Y"
• Transform defensive tone: "We had to rollback because..." → "Rolled back to [previous version] due to [root cause]"
• Preserve urgency markers and severity indicators (needed for status classification)
• Keep technical terms intact (oversimplification loses information; non-technical audiences still need accuracy)
• Maintain causal chains and specific metrics (specific > generic)

Step 3: Status classification

Apply criteria consistently (inconsistency confuses stakeholders and erodes trust):

• GREEN: Fully complete with no follow-up, all verification done
• YELLOW: Resolved with follow-up needed, blocked on dependencies, partial completion
• RED: Active critical issues, production impact, urgent intervention needed

Always document reasoning: "Status: YELLOW (deployment successful but monitoring pending)" not just "Status: YELLOW"

Step 4: Action item specificity

Vague action items cannot be executed. Every next step MUST include:

• Specific action verb (investigate, deploy, coordinate, document) — "fix" is too vague
• Clear scope (what exactly needs doing) — define the boundary
• Ownership implication (who or which team) — someone must be accountable
• Timeline marker when available (IMMEDIATE, by EOW, this sprint) — explicit > implied

Gate: Output follows template structure with professional tone and all specificity rules applied. Proceed only when gate passes.

Phase 4: VERIFY

Goal: Confirm transformation quality before delivery. All gates must pass; proceed only when complete.

Step 1: Compare output against extracted propositions — NO information loss allowed. If a fact from Phase 1 doesn't appear in output, it belongs in Technical Details.

Step 2: Verify technical accuracy — terms, metrics, causal chains preserved exactly. Preserve exact technical terms ("database issues" for "Redis cluster failover") — specificity is required.

Step 3: Confirm status indicator matches actual severity. Check reasoning against actual criteria (GREEN ≠ YELLOW vs YELLOW ≠ RED boundaries).

Step 4: Validate action items are specific — check each next step for (verb, scope, owner, timeline). "Fix the issue" fails; "Complete Redis failover testing in staging (DevOps, by EOW)" passes.

Step 5: Check appropriate detail level for target audience. If audience is non-technical, Technical Details should bridge jargon with plain explanations without losing precision.

Step 6: Document transformation summary to prove gate passage:

## Transformation Summary
Input type: [type]
Propositions extracted: [N]
Status assigned: [GREEN|YELLOW|RED] ([reasoning])
Information loss: None
Template applied: standard

Gate: All verification checks pass. Transformation is complete. Complete all 6 steps before delivering.

Examples

Example 1: Multi-Propositional Sentence

User says: "I fixed the database issue but then the API started failing so I had to rollback and now we're investigating the connection pool settings which might be related to the recent Kubernetes upgrade."

Actions:

• Extract 5 propositions: DB fix, API failure, rollback, pool investigation, K8s link (PARSE)
• Categorize: Status=rollback done, Blockers=pool+K8s, Actions=investigating (STRUCTURE)
• Apply template with YELLOW status, specific next steps (TRANSFORM)
• Verify no facts lost, technical terms preserved (VERIFY)

Result: Structured update with clear status and actionable next steps

Example 2: Defensive Blocker Communication

User says: "I can't make progress because the API team hasn't responded in 3 days and my sprint is at risk"

Actions:

• Extract urgency, duration, dependency, impact propositions (PARSE)
• Categorize: Blocker=API spec, Impact=sprint risk, Timeline=3 days (STRUCTURE)
• Apply template with YELLOW status, escalation-focused next steps (TRANSFORM)
• Verify urgency preserved, defensive tone neutralized (VERIFY)

Result: Neutral status report with clear escalation path

Example 3: Crisis Communication

User says: "The latest deploy broke checkout completely, users are getting 500 errors, we rolled back but some orders might be lost"

Actions:

• Extract severity, system affected, user impact, rollback status, data risk (PARSE)
• Categorize: Status=rolled back, Impact=orders lost, Blocker=data recovery (STRUCTURE)
• Apply template with RED status, IMMEDIATE/URGENT tiered next steps (TRANSFORM)
• Verify crisis severity reflected, no false reassurance in tone (VERIFY)

Result: RED status report with tiered emergency response actions

Error Handling

Error: "Missing Context in Input"

Cause: Technical terms or acronyms critical to business impact are undefined.

Solution:

• Ask user for clarification on terms critical to status classification — speculation causes wrong status assignments
• Make reasonable inferences only for minor details; flag all assumptions explicitly in Technical Details section
• Complete transformation while waiting — provide output with a note: "Status classification assumed X because Y was undefined"

Error: "Ambiguous Status Classification"

Cause: Input contains mixed signals (e.g., issue resolved but monitoring incomplete).

Solution:

• Default to YELLOW when unclear between GREEN/YELLOW — YELLOW preserves urgency for follow-up without false reassurance
• Default to RED only with clear critical indicators: production impact (users affected), data loss (unrecoverable), or ongoing crisis (not yet mitigated)
• Document reasoning in parenthetical: "Status: YELLOW (deployment successful but monitoring pending)" — transparency prevents misinterpretation

Error: "Multi-Thread Update Contamination"

Cause: Input contains multiple unrelated topics that could cross-contaminate status classifications.

Solution:

• Process each thread as separate proposition set (Phase 1 extraction per thread)
• Apply template independently to each thread (Phase 2-3 per thread)
• Combine with clear thread identification in final output (use headers: "Thread A: Deployment", "Thread B: Data Recovery")
• Ensure status indicators are thread-specific (Thread A may be GREEN while Thread B is RED) — separate outcomes, separate classifications

Error Handling Principles

Constraint distribution in error handling:

• Summarizing before extracting = loses facts. Complete Phase 1 fully before proceeding.
• Status is "obvious" = assumption. Apply classification criteria consistently, document reasoning.
• Technical details not needed for non-technical audience = false. Always include Technical Details; bridge jargon with explanation.
• Action items are implied = stakeholders cannot execute implied work. Write explicit (verb, scope, owner, timeline) for every next step.
• Professional tone is "close enough" = defensive language still embedded. Apply ALL transformation rules: hedging → direct, emotional → neutral, vague → specific.

References

Reference Files

• ${CLAUDE_SKILL_DIR}/references/templates.md: Status-specific templates, section formats, phrase transformations
• ${CLAUDE_SKILL_DIR}/references/examples.md: Complete transformation examples with proposition extraction