alivirgo
- 392 skills
- 0 followers
- 1 day ago last updated
- ▌ Draft Meeting Agendas · alivirgoHow to generate high-efficiency, time-boxed meeting agendas with explicit decision criteria, pre-reads, and discussion ownership.
- ▌ Early Exit Evaluation · alivirgoHow autonomous debugging agents terminate exploratory searches immediately upon locating the first conclusive root-cause bug, cutting troubleshooting tokens by 60%.
- ▌ Abbreviated Git Status · alivirgoHow autonomous coding agents use short-format and porcelain git status flags to eliminate 85% of shell output token overhead while preserving exact branch and staging state.
- ▌ Artifact Reuse Pattern · alivirgoHow autonomous agents maintain single living state documents (implementation plans, walkthroughs) through in-place mutations rather than generating duplicate ephemeral files.
- ▌ Ask AI To Audit Itself · alivirgoLeverage the Reflexion and Critic-Actor prompting patterns to force LLMs to self-correct logic errors, overlooked edge cases, and factual slips before final delivery.
- ▌ Ask Short Answer First · alivirgoMaster the Bottom Line Up Front (BLUF) prompting pattern to get the direct 2-line answer immediately, eliminating preamble fluff and saving reading time.
- ▌ Boolean Flag Responses · alivirgoHow to constrain diagnostic checks, compliance audits, and triage gates to single-token boolean/enum responses (TRUE/FALSE), cutting output tokens by 99%.
- ▌ Compact Error Handling · alivirgoHow to eliminate verbose, repetitive try-catch blocks in generated code by using multi-exception tuples, contextlib.suppress, and functional Result types, cutting error boilerplate by 60%.
- ▌ Compact Status Updates · alivirgoHow to use concise, structured tool metadata (toolAction, toolSummary) and 1-sentence milestone broadcasts instead of verbose conversational progress monologues, saving 90% of intermediary tokens.
- ▌ Compressed JSON Repair · alivirgoHow agent runtimes fix malformed LLM JSON outputs locally using deterministic parsing heuristics (jsonrepair, regex) rather than wasting expensive re-prompting roundtrips.
- ▌ Head Tail Log Clipping · alivirgoHow autonomous agents truncate massive terminal and build logs by retaining the first 20 lines (head) and last 40 lines (tail) while eliding the middle, eliminating 90% of log token bloat.
- ▌ Highlight Action Items · alivirgoHow to command AI to top-load next steps and action items at the very beginning of responses to ensure zero dropped tasks and instant execution.
- ▌ In Place File Mutation · alivirgoWhy autonomous coding agents must modify source files directly in-place using replace_file_content rather than creating temporary sidecar files (temp.py, new.ts), eliminating filesystem clutter and multi-turn cleanup overhead.
- ▌ Inline Type Signatures · alivirgoHow to enforce native language type hints (TypeScript syntax, Python PEP 484) while stripping duplicate JSDoc/Sphinx @param docstrings, eliminating 40% of signature token bloat.
- ▌ JSON Minified Payloads · alivirgoHow agent runtimes and tool callers serialize JSON payloads with compact single-line formatting (separators=(',', ':')), eliminating 40% of indentation whitespace tokens.
- ▌ Minimal Diff Formatter · alivirgoHow to format code changes using standard unified diffs (```diff with + and - lines) instead of full-file code blocks, cutting output token generation by 85% and improving developer review velocity.
- ▌ Outline Before Writing · alivirgoHow to use the Structural Outline Gate to lock in logical progression, prevent rambling drafts, and eliminate 80% of document rewriting.
- ▌ Scratch File Isolation · alivirgoWhy all temporary debug scripts, one-off AST refactorers, and test payloads must be quarantined inside a dedicated scratch/ directory, eliminating git status noise and test runner collisions.
- ▌ System Prompt Minifier · alivirgoHow to build and run an automated system prompt minifier that strips markdown comments, compresses redundant whitespace, and rewrites polite fluff into dense imperative grammar, cutting prompt size by 50%.
- ▌ Tell AI What Not To Do · alivirgoHow to use explicit Negative Constraints ('Do NOT use third-party libraries', 'Do NOT use passive voice') to prune hallucination paths and lock in exact specifications.
- ▌ Tell AI When In A Rush · alivirgoHow to trigger Emergency High-Velocity Mode for instant 3-second answers during live client meetings, outages, and negotiation standoffs.
- ▌ Token Budget Allocator · alivirgoHow agent runtimes enforce mathematical token quotas across system instructions, tool schemas, RAG context, and conversation history, preventing single-source context exhaustion.
- ▌ Type Alias Compression · alivirgoHow to extract repetitive nested generic signatures into clean top-level type aliases (type APIRes<T> = Promise<Result<T, AppError>>), cutting TypeScript/Python type annotation tokens by 45%.
- ▌ Act As A Personal Tutor · alivirgoTransform any AI assistant into an interactive, step-by-step personal tutor that assesses prior knowledge, explains concepts in bite-sized chunks, and quizzes comprehension before advancing.
- ▌ Ast Refactoring Scripts · alivirgoHow autonomous agents generate and execute local Abstract Syntax Tree (AST) scripts (libcst, ts-morph, jscodeshift) for batch codebase migrations, eliminating 99% of token costs.
- ▌ Binary File Suppression · alivirgoHow autonomous coding agents detect and suppress raw binary, compiled, and media files from text context ingestion, preventing catastrophic context corruption and token overflow.
- ▌ Concise Output Enforcer · alivirgoHow to enforce strict negative output constraints to eliminate conversational preambles, post-implementation monologues, and discursive filler, slashing response tokens by 60%.
- ▌ Exit Code Fast Checking · alivirgoHow to evaluate CLI command success directly via process exit codes ($? == 0), suppressing thousands of lines of successful build/compilation stdout logs.
- ▌ Give AI Direct Feedback · alivirgoHow to use the 3-Part Direct Feedback Formula to steer AI outputs without vague trial-and-error or starting chats from scratch.
- ▌ Inline Lambda Functions · alivirgoHow to eliminate verbose multi-line helper boilerplate by using concise inline lambda expressions and arrow functions, cutting callback token generation by 55%.
- ▌ Lazy Dependency Loading · alivirgoHow autonomous coding agents rely on parametric knowledge first and fetch external SDK documentation Just-In-Time (JIT) only upon compiler or import failures, cutting upfront context bloat by 85%.
- ▌ No Preamble Code Blocks · alivirgoHow to enforce zero introductory throat-clearing ('Here is the code:') so responses start strictly on Line 1 with the code block or raw script, enabling direct CLI piping and saving output tokens.
- ▌ Output Final Draft Only · alivirgoHow to use strict negative formatting constraints to suppress conversational chit-chat ('Sure, here is...', 'Hope this helps!') for 1-click copy-pasting and script automation.
- ▌ Semantic Key Shortening · alivirgoHow to use standardized, universally understood abbreviations (src, dst, err, req, res, fn, qty) in JSON schemas and structured outputs to cut payload token consumption by 55%.
- ▌ Summarized RAG Memories · alivirgoHow to compress retrieved vector/RAG document chunks into 3-bullet atomic facts before injecting them into frontier LLM context, cutting RAG token bloat by 85%.
- ▌ Token Bounded Summaries · alivirgoHow to constrain operational, PR, and milestone summaries to a strict 50-token ceiling ([ACTION] -> [CHANGE] -> [STATUS]), eliminating 400-word narrative essays and cutting output tokens by 88%.
- ▌ Add Your Own Human Voice · alivirgoTechniques and prompt frameworks to eliminate robotic AI clichés, infuse authentic personal voice, and blend human perspective with AI synthesis.
- ▌ Ask AI Confidence Rating · alivirgoPrompt patterns to extract calibrated confidence scores, identify hidden assumptions, and flag low-certainty claims before making critical decisions.
- ▌ Ask AI To Check Its Math · alivirgoTechniques to eliminate arithmetic hallucinations by enforcing Chain-of-Thought scratchpads, explicit order-of-operations steps, and programmatic Python execution.
- ▌ Ask What Info Is Missing · alivirgoHow to use Context Gap Discovery and the Interview-Me pattern to stop AI from hallucinating missing variables and uncover blind spots.
- ▌ Batch API Job Processing · alivirgoHow to route non-interactive workloads (evaluations, mass refactoring audits, dataset generation) through Anthropic and OpenAI Asynchronous Batch APIs to cut token bills 50%.
- ▌ Compact YAML Frontmatter · alivirgoHow to design lean, high-density YAML frontmatter for documentation, skills, and RAG knowledge bases, eliminating 70% of header token bloat.
- ▌ Cross Check Legal Advice · alivirgoHow to safely use AI for legal issue-spotting, contract clause translation, and preparing targeted questions for licensed legal counsel.
- ▌ Dense Shorthand Notation · alivirgoHow to use high-density symbolic operators (->, =>, |, ::, !) to compress complex business logic, state machines, and system rules, reducing prompt tokens by 65%.
- ▌ Flag Missing Facts First · alivirgoHow to command AI to conduct a Pre-Flight Information Audit and request missing data before attempting to solve complex problems.
- ▌ Huggingface Transformers · alivirgoOperational skill for Hugging Face Transformers: pipelines, tokenizers, fine-tuning, and inference with AutoModel APIs.
- ▌ Keep Instructions Direct · alivirgoWhy leading with precise imperative action verbs ('Refactor', 'Audit', 'Synthesize') maximizes model attention and eliminates preamble fluff.
- ▌ Lint Auto Fix Delegation · alivirgoHow autonomous coding agents delegate formatting, unused imports, and styling rules to local linters (ruff, biome, eslint --fix) rather than wasting LLM tokens on manual syntax fixes.
- ▌ Max Tokens Ceiling Guard · alivirgoHow to clamp max_tokens parameter to task-specific budgets (1 to 300 tokens) rather than defaulting to 4,096, preventing runaway generation loops and capping financial exposure.
- ▌ Proofread Before Sending · alivirgoHow to use AI as an objective Emotional Tone & Subtext Auditor to catch unintended passive-aggression, defensiveness, or submissive phrasing before hitting send.
- ▌ Remember AI Cutoff Dates · alivirgoHow to prevent obsolete code, deprecated APIs, and outdated tax advice by enforcing Temporal Grounding and live web search triggers.
- ▌ Selective Grep Filtering · alivirgoHow to use precise filetype globs (Includes: ['src/**/*.ts', '!**/*.test.ts']) and literal string matching to eliminate 90% of search noise and false-positive matches.
- ▌ Set A Word Count Ceiling · alivirgoHow to enforce hard word ceilings ('under 60 words', 'max 3 sentences') to eliminate fluff, boost readability, and slash token costs.
- ▌ Skip Polite Filler Words · alivirgoWhy stripping polite filler words ('Please', 'Hello', 'Could you kindly') saves input tokens, eliminates conversational skew, and sharpens response quality.
- ▌ Stop Sequence Truncation · alivirgoHow to configure custom stop sequences (stop: ['\\n\\n', '```\\n', '---']) at the inference sampler layer to terminate generation instantly upon deliverable completion, cutting trailing tokens by 50%.
- ▌ Stop Wrong Answers Early · alivirgoWhy hitting 'Stop Generating' within 3 seconds of a misdirected response saves tokens, eliminates context poisoning, and speeds up workflow.
- ▌ Symbol Declaration Index · alivirgoHow to generate and query compact 1-line symbol indexes (tags / symbols.tsv) using Tree-sitter and Universal Ctags, enabling instant symbol localization without multi-file text scans.
- ▌ Tell AI Your Time Budget · alivirgoHow to anchor prompts with explicit execution time limits ('I have 10 minutes before a demo' vs 'I have 2 days') to get calibrated solutions.
- ▌ Token Efficient Markdown · alivirgoHow to format markdown documentation, tables, and artifacts with high structural density, eliminating decorative ASCII dividers, padded table spaces, and multi-newline whitespace bloat.
- ▌ Use Plain Text Over Pdfs · alivirgoWhy extracting and pasting clean plain text or Markdown beats raw PDF uploads by eliminating OCR latency, visual token bloat (up to 80% savings), and table scrambling.
- ▌ Zero Boilerplate Imports · alivirgoHow to eliminate redundant import blocks when displaying localized code patches and diffs, relying on existing file imports and automated CLI import sorters.
- ▌ Zero Cat Bash Prevention · alivirgoWhy autonomous agents must use structured native file viewing tools (view_file) rather than terminal shell commands (cat, type, Get-Content), enabling 1-indexed line numbering and slice bounding.
- ▌ Ask How To Improve Prompt · alivirgoTurn the AI into an expert Prompt Engineer to critique your instructions, fill missing context gaps, and generate reusable high-performance prompt templates.
- ▌ Compressed System Prompts · alivirgoHow to refactor wordy, repetitive system prompts into high-density imperative token structures, cutting baseline system prompt costs by 70% while improving instruction adherence.
- ▌ Context Pruning Directive · alivirgoHow agent runtimes dynamically prune stale tool outputs, dead build logs, and historical turns to eliminate quadratic token growth ($O(N^2)$) and prevent attention degradation.
- ▌ Deduplicated File Caching · alivirgoHow agent runtimes intercept redundant file reads within the same conversation trajectory using mtime and SHA-256 hash checks, eliminating 40% of duplicate context ingestion tokens.
- ▌ Draft Multiple Variations · alivirgoHow to prompt for high-contrast variations using psychological angles (Curiosity, Direct Benefit, Urgency, Counter-Intuitive) rather than synonym swaps.
- ▌ Give AI Situation Context · alivirgoHow to anchor prompts with the 4-Pillar Situation Framework (Role, Scale, Budget, Constraints) to stop AI from over-engineering or giving irrelevant advice.
- ▌ Line Bounded File Reading · alivirgoHow autonomous coding agents use StartLine and EndLine parameters to read targeted 30-to-50 line slices rather than unbounded 2,000-line files, eliminating 90% of file ingestion tokens.
- ▌ Logit Bias Token Steering · alivirgoHow to apply logit_bias parameters at the transformer sampling layer to mathematically force 1-token boolean/enum decisions and ban conversational filler words, cutting output tokens by 99%.
- ▌ Minimal Artifact Metadata · alivirgoHow to structure high-density ArtifactMetadata fields (Summary, UserFacing, RequestFeedback) during artifact creation and editing, cutting metadata overhead by 80%.
- ▌ No Redundant Type Casting · alivirgoHow to eliminate redundant type assertions (as unknown as T) and defensive coercion wrapping (str(str(x)), Boolean(y === true)), cutting code generation token bloat by 30%.
- ▌ No Vendor Directory Scans · alivirgoHow agent tools and recursive search crawlers enforce strict glob exclusion on vendor directories (node_modules, venv, target, dist, .git), preventing catastrophic 80,000-token context floods.
- ▌ One Shot Subagent Prompts · alivirgoHow orchestrator agents formulate comprehensive, self-contained task prompts for child subagents (Task, Boundaries, Stop Conditions, Return Schema), eliminating 75% of multi-turn clarification ping-pongs.
- ▌ Reactive Wakeup Signaling · alivirgoWhy autonomous agents must rely on event-driven runtime wakeups instead of polling status loops (manage_task status) while waiting for background jobs, eliminating 95% of waiting token waste.
- ▌ Ripgrep Over Grep Command · alivirgoWhy autonomous agents must use structured native code search tools (grep_search) rather than terminal shell commands (grep -rn), eliminating ANSI formatting noise and capping match overflows.
- ▌ Simple Understanding Quiz · alivirgoHow to use the Testing Effect and Active Retrieval practice to diagnose knowledge gaps, eliminate illusions of competence, and lock in concepts.
- ▌ Single Pass Multi Replace · alivirgoHow to edit multiple separated blocks across a single file in one atomic turn using multi_replace_file_content, eliminating line-drift offset errors and cutting edit roundtrips by 70%.
- ▌ Targeted Line Range Edits · alivirgoHow to apply precise line-bounded code edits using replace_file_content ([StartLine, EndLine] windowing) rather than rewriting complete 500-line files, cutting edit output tokens by 95%.
- ▌ Write Step By Step How To · alivirgoHow to generate bulletproof Standard Operating Procedures (SOPs) and technical tutorials with exact click paths, expected visual results, and failure troubleshooting.
- ▌ Avoid Rephrasing Good Text · alivirgoHow to perform surgical edits on awkward sentences without destroying your authentic voice, personal humor, or working copy.
- ▌ Bullet Point Summary First · alivirgoHow to extract high-yield, action-oriented bullet summaries from long documents, meetings, and articles using the 3-Bullet Digest framework.
- ▌ Cached System Instructions · alivirgoHow to structure hierarchical, immutable system instructions to maximize KV-cache hits on Anthropic and OpenAI, slashing Time-to-First-Token (TTFT) by 80% and input costs by 90%.
- ▌ Incremental Diff Ingestion · alivirgoHow autonomous agents use compact unified git diffs (git diff -U3, git diff @{u}..HEAD) instead of full-file ingestion for PR reviews and commit audits, reducing review tokens by 90%.
- ▌ Lm Studio Local Offloading · alivirgoHow autonomous engineering systems route mechanical sub-tasks (git commits, log triage, unit test boilerplate) to local GGUF models via LM Studio/Ollama, cutting cloud API spend by 60%.
- ▌ No Polite Chitchat Closing · alivirgoHow to eliminate conversational closing signatures ('Hope this helps!', 'Let me know if you have questions!') to save output tokens and terminate turns immediately upon deliverable completion.
- ▌ Openapi Endpoint Filtering · alivirgoHow autonomous agents slice monolithic 10,000-line OpenAPI/Swagger specifications down to operation-targeted sub-schemas, eliminating 99% of API specification token bloat.
- ▌ Verify Names And Locations · alivirgoWhy proper nouns, addresses, and job titles are the #1 source of AI hallucinations, and how to use the 4-Point Entity Verification Ladder to prevent embarrassing mistakes.
- ▌ Compare Two Products Fairly · alivirgoHow to use Symmetrical Head-to-Head prompting to compare software, hardware, or tools objectively with transparent pricing and fatal flaw analysis.
- ▌ Fill In Missing Assumptions · alivirgoHow to use Assumption Surfacing and Sensitivity Analysis to uncover silent presuppositions, prevent architectural failure, and stress-test recommendations.
- ▌ Flag Speculative Statements · alivirgoHow to use Epistemic Tagging ([FACT], [HEURISTIC], [SPECULATION]) to force AI to visibly separate verified truths from estimates and predictions.
- ▌ Gitignore Respecting Search · alivirgoHow autonomous agents use gitignore-aware search engines (ripgrep) to automatically exclude node_modules, build artifacts, and vendor directories, eliminating 95% of search noise.
- ▌ Header Only C Cpp Ingestion · alivirgoHow autonomous agents inspect .h/.hpp header files first to understand class contracts and struct layouts before reading heavy .cpp implementation files, slashing C++ context token spend by 85%.
- ▌ Multi Tool Batch Invocation · alivirgoHow autonomous agents emit multiple independent tool calls in a single turn (tool_calls array), enabling parallel execution via Promise.all and slashing turn roundtrips by 70%.
- ▌ One Sentence Elevator Pitch · alivirgoHow to force AI to condense complex apps, startups, and initiatives into a crystal-clear, 1-sentence elevator pitch using the XY Positioning Formula.
- ▌ Regex One Liner Refactoring · alivirgoHow to replace 20-line procedural character-by-character parsing loops with concise, compiled Regular Expressions, reducing string manipulation code size by 75%.
- ▌ Single Line Error Summaries · alivirgoHow to condense 100-line stack traces into structured single-line root cause statements (🚨 [FAIL: file:line] ErrorType: Message), slashing error reporting tokens by 95%.
- ▌ Skip Explaining The Obvious · alivirgoHow to suppress introductory 101 definitions and force AI to assume senior domain competence, saving tokens and reading time.
- ▌ Standard Library Preference · alivirgoHow to leverage modern built-in standard library utilities (functools, structuredClone, crypto.randomUUID, itertools) instead of writing custom helper boilerplate or pulling in heavy third-party dependencies.
- ▌ Start Fresh Chat New Topics · alivirgoWhy long-running mega-threads cause context pollution, latency spikes, and hallucinations, and how to use the State Handoff Pattern to reset cleanly.