Skip to content

DICTATION.md

Latest commit

 

History

History
330 lines (318 loc) · 5.83 KB

DICTATION.md

File metadata and controls

330 lines (318 loc) · 5.83 KB
name dictation
description Dictation instructions for fixing speech-to-text errors and improving text quality in gh-aw workflows

Dictation Instructions

Technical Context

gh-aw (GitHub Agentic Workflows) is a CLI extension for GitHub that compiles markdown workflow files into GitHub Actions YAML. It enables AI-powered workflows using natural language instructions with support for multiple engines (Copilot, Claude, Codex, Gemini), tools (GitHub API, bash, web-fetch, playwright), and security features (safe-outputs, network permissions, integrity levels).

Project Glossary

AGENTS.md CLAUDE.md COPILOT_GITHUB_TOKEN COPILOT_MODEL DOCKER_HOST GH_AW_DEFAULT_DETECTION_MODEL GH_AW_DEFAULT_MAX_EFFECTIVE_TOKENS GH_HOST GITHUB_TOKEN MCP_GATEWAY_SESSION_TIMEOUT README.md YAML acceptEdits actions activation actor add-comment add-wizard agentic agentic-workflows agent agent-output.json allowed allowed-domains allowed-files allowed-github-references allowed-repos api-target append approval-labels approved artifact artifacts audit avoidHourBoundary avoidPeakMinutes aw-info.json aw.json aw.yml bash blocked blocked-users body branch bun bypassPermissions cache cache-memory call-workflow checkout client-id codex comment-memory compile compilation concurrency conclusion contents copilot create-discussion create-issue create-pull-request create-pull-request-review-comment cron custom custom-safe-outputs daily default defaults deno dependencies detection dev-tools discussion discussion-number discussions dispatch-workflow domain download-artifact edit effective-tokens end-date endpoint engine engine.bare engine.env engine.mcp engine.model engine.permission-mode env ephemerals experiment experiments expires fallback-to-issue feature-flags field_node_id firewall firewall-audit-logs footer frontmatter fuzzy-schedule gemini gemini-flash gemini-pro gh-aw github github-app github-token go.mod gpt-5 gpt-5-mini haiku headers hourly http id-token ignore ignore-if-missing import-schema imports inlined-imports integrity integrity-reactions issue issue-comment issue-number issues job jobs keepaliveInterval label-command labels large lock.yml lockfile manifest markdown max-continuations max-effective-tokens max-patch-size max-runs max-turns mcp mcp-gateway mcp-scripts mcp-servers member merge merged midnight min-integrity min-samples mini model model-alias network network.allowed node noop noon notify observability opentelemetry opus organization otel otlp owner package.json packages payloadDir pending permissions playwright post-steps pre-activation pre-steps prepend private-key projected-effective-tokens prompt.txt protected-files pull-request pull-request-number pull-request-target pull-requests pull_request pull_request_target push push-to-pull-request-branch python rate-limiting redirect refusal-labels registry replace repo repo-memory report-incomplete repository repository-dispatch required-labels required-title-prefix retention-days runs-on runs-on-slim runtime runtimes safe-outputs sandbox schedule security serena sessionTimeout skip-if-match slash-command sonnet spanId staged start-date state.json state.runs steps.sanitized.outputs.body steps.sanitized.outputs.text steps.sanitized.outputs.title stop-after target-repo threat-detection timeout-minutes timezone title-prefix toolsets tools tools.github tools.timeout total-effective-tokens traceId traceparent trusted-users trustedBots unapproved update-issue update-project upload-artifact upload-asset user-rate-limit uv variants version warn web-fetch web-search weekly weight workflow-call workflow-dispatch workflow_call workflow_dispatch workflows

Fix Speech-to-Text Errors

Common misrecognitions to correct:

  • "GH away" → "gh-aw"
  • "G H A W" → "gh-aw"
  • "lock Y M L" → "lock.yml"
  • "Y A M L" → "YAML"
  • "MCP" → "MCP" (not "M C P")
  • "front matter" → "frontmatter" (one word)
  • "safe outputs" → "safe-outputs" (hyphenated)
  • "workflow dispatch" → "workflow-dispatch" (hyphenated in YAML context)
  • "pull request" → "pull-request" (hyphenated in YAML context)
  • "cop pilot" → "copilot" (one word)
  • "code X" → "codex" (one word)
  • "bypass permissions" → "bypassPermissions" (camelCase)
  • "accept edits" → "acceptEdits" (camelCase)
  • "Gemini" → "gemini" (lowercase in YAML context)
  • "engine dot model" → "engine.model"
  • "engine dot env" → "engine.env"
  • "engine dot MCP" → "engine.mcp"
  • "tools dot GitHub" → "tools.github"
  • "MCP gateway" → "mcp-gateway" (hyphenated)
  • "MCP scripts" → "mcp-scripts" (hyphenated)
  • "cache memory" → "cache-memory" (hyphenated)
  • "repo memory" → "repo-memory" (hyphenated)
  • "allowed repos" → "allowed-repos" (hyphenated in YAML context)
  • "min integrity" → "min-integrity" (hyphenated in YAML context)
  • "pre activation" → "pre-activation" (hyphenated)
  • "A W dot Y M L" → "aw.yml"
  • "A W dot JSON" → "aw.json"
  • "GitHub token" → "github-token" (hyphenated in YAML context)
  • "GitHub app" → "github-app" (hyphenated in YAML context)
  • "steps dot sanitized dot outputs dot text" → "steps.sanitized.outputs.text"
  • "fuzzy schedule" → "fuzzy-schedule" (hyphenated)

Clean Up and Improve Text

  • Remove filler words: humm, you know, um, uh, like, basically, actually, so, well, right, okay
  • Remove false starts and repeated words
  • Improve clarity and sentence structure
  • Make text more professional and concise
  • Fix run-on sentences
  • Correct grammar and punctuation
  • Maintain the user's intended meaning and tone
  • Preserve technical terminology exactly as provided

Guidelines

You do not have enough background information to plan or provide code examples.

  • do NOT generate code examples
  • do NOT plan steps
  • focus on fixing speech-to-text errors and improving text quality
  • remove filler words (humm, you know, um, uh, like, basically, actually, etc.)
  • improve clarity and make text more professional
  • maintain the user's intended meaning