Technical Writer

technical-writer

Purpose: Transform rough technical notes or verbose AI-generated drafts into clear, scannable technical documentation.

Input Schema

| Field | Type | Required | |---|---|---| | doc_type | string | yes - readme, runbook, adr, api-doc, design-doc | | content | string | yes - raw notes or draft | | preserve_facts | string[] | yes | | audience | string | yes - e.g., "on-call engineers", "external API consumers" | | tone | string | yes |

{
  "doc_type": "runbook",
  "content": "When the service goes down you should check the logs and then maybe restart it. The service is very important and serves as a critical component in our infrastructure.",
  "preserve_facts": ["service name: payment-processor", "restart command: systemctl restart payment-processor"],
  "audience": "on-call engineers",
  "tone": "direct"
}

Output Schema

{
  "title": "Runbook: payment-processor",
  "sections": {
    "symptoms": "...",
    "diagnosis": "...",
    "resolution": "...",
    "escalation": "..."
  },
  "patterns_removed": [1, 7, 8, 14],
  "word_count": 120
}

Prompt Flow

Pass 1: Rewrite into appropriate doc structure for doc_type. Remove: significance inflation (P1), promotional adjectives (P4), AI vocab (P7), copula avoidance (P8), boldface overuse (P14), inline-header lists (P15), filler (P22). Preserve preserve_facts.

Pass 2 Audit + Final: Check for any remaining vague attributions (P5) or -ing tail clauses (P3). Rewrite to specific and direct.

Examples

Short - runbook symptom

Before: "The service serves as a critical component and may be experiencing issues that could potentially be causing downstream impact." After: "Symptom: payment-processor is returning 5xx. Downstream: checkout is blocked."

Medium - README section

Before: "This groundbreaking tool boasts the ability to leverage cutting-edge algorithms to enhance developer productivity." After: "This tool reduces CI build time by parallelizing your test suite. It requires Python 3.11+ and a Redis instance."

Long - ADR

Before: "We have decided to utilize a microservices architecture, underscoring its vital role in our evolving landscape and highlighting its significance for long-term scalability, cultivating a foundation for future growth." After: "Decision: adopt a microservices architecture for the billing domain.\nContext: the monolith's deploy cycle is 2 hours; billing changes block all other deploys.\nConsequences: services will communicate over gRPC; the team needs observability tooling."

Unit Tests

# tests/skills/test_technical_writer.py
from ai_pattern_scrubber import detect_patterns

SHORT = "Symptom: payment-processor is returning 5xx. Downstream: checkout is blocked."
MEDIUM = "This tool reduces CI build time by parallelizing your test suite. It requires Python 3.11+ and a Redis instance."

def test_tech_short_no_high_severity():
    assert not [h for h in detect_patterns(SHORT) if h.severity == "high"]

def test_tech_medium_no_promotional_language():
    assert not [h for h in detect_patterns(MEDIUM) if h.id == 4]

def test_tech_medium_no_ai_vocab():
    assert not [h for h in detect_patterns(MEDIUM) if h.id == 7]

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: nicholashidalgo
• Source: nicholashidalgo/claude-skillforge
• License: MIT

Install and usage instructions live in the source repository linked above.