synthesisengineering

Post-implementation verification protocol that catches incomplete work, hidden shortcuts, and gaps between 'tests pass' and 'production works.' Systematically traces data chains, detects placeholders, audits test honesty, and verifies environment parity. Use when asked to: verify implementation, check completeness, implementation review, is this done, built properly, no shortcuts, verify work, completion check, integrity check, self-review, pre-PR check, ship check.

Lightweight project management system designed for human-agent collaboration, optimized for context preservation across sessions. Use when asked to: project management, project setup, project tracking, synthesis project, manage project, set up project, project structure, session protocol.

Validate, convert, and author content for Google's Open Knowledge Format (OKF v0.1) — the markdown-plus-YAML-frontmatter knowledge-bundle spec announced 2026-06-12. Includes a conformance validator (the checker Google's own OKF repo ships none of), a converter that backfills OKF frontmatter onto an existing markdown corpus idempotently, and the proven repo-by-repo conversion procedure. Use when adopting OKF for an "LLM wiki" style knowledge base, auditing conformance, or converting a corpus of markdown notes/docs into a conformant, agent-readable bundle.

Discipline for catching the lazy-shortcut antipattern in AI-assistant output. Makes the costume vocabulary explicit so agents recognize when their drafts have slid into deferral, dismissal, or false consultation. Includes the constraint-first protocol, sub-agent dispatch and acceptance hygiene, and a pre-response self-check. Use when asked to: avoid shortcuts, audit for laziness, check for deferral, enforce best solution, no shortcuts, anti-shortcut, constraint-first, sub-agent hygiene, costume vocabulary.

Cross-provider model-tier convention for agentic work: three role labels (judgment, routine, bulk — formerly frontier, efficient, light) resolved to current model IDs per provider in tiers.yaml, so skills, project docs, and memory never hardcode model names. Use when asked about: model tiers, which model, model selection, judgment model, routine model, bulk model, frontier model, efficient model, switch models, model equivalents across providers, update model table.

Fetch AI-generated meeting notes and full transcripts (e.g., Google Meet + Gemini) from the user's Gmail/Drive into local markdown files. Tool-agnostic: works with any Gmail + Drive MCP (Anthropic hosted connectors, self-hosted workspace-mcp, or others). Replaces the manual email → Google Doc → export-markdown → Downloads → move workflow. Use when asked to: fetch meeting transcript, pull standup, grab meeting notes, sync meetings, download transcript, get Gemini notes, import meeting.

Reconcile pending, missed, and incomplete commitments after any gap in the daily-ritual cadence (travel, family visits, crunch weeks). Sweeps daily plans, transcripts, and project context over an arbitrary window; classifies every surfaced item (including expired-for-learning); produces a dated catch-up ledger document; routes survivors without flooding the daily plan. Use when asked to: catch up on missed tasks, sweep pending items, catch-up ledger, reconcile backlog, what did I miss, post-vacation catch-up, falling behind recovery.

Day-start and day-end checklists for synthesis engineering projects. Execute dependency-ordered rituals for context optimization, Slack sync, catch-up reads, PR reviews, day planning, and communications. Use when asked about: daily ritual, morning routine, day start, day end, daily checklist, morning checklist, end of day checklist, daily workflow.

Produce a synthesis report from 2+ workspaces with explicit per-workspace citations and a visible audit trail. Use when asked to: synthesize, compare workspaces, cross-reference, multi-workspace research, cite from multiple workspaces.

Score the agent's previous turn against an explicit rubric covering groundedness, completeness, safety, and format. Surface failures the agent can fix before the user has to flag them. Use when asked to: self review, audit your response, check your previous answer, score this turn, was that grounded, did you cover everything, post-response review.

Produce a first draft, then revise it against a rubric until each rubric item is satisfied. Two-stage prompt that separates generation from criticism so each gets full attention. Use when asked to: draft, write, compose, revise, improve a draft, iterate on writing, refine, polish, edit my draft, rewrite, second pass.

Extract factual claims from a text, retrieve supporting context for each, and assign a verdict of supported, contradicted, or unknown. Use when asked to: fact check, verify claims, check accuracy, audit for errors, validate the assertions, ground-truth the draft, find unsupported claims.

Produce a multi-style summary of a document: an executive summary for skim readers, a key-points list for reference, and an action-items list for owners. Three output sections in a fixed structured format. Use when asked to: summarize, tldr, give me the key points, what does this say, executive summary, action items from, condense, distill.

Search the active workspace's three-tier memory (recent turns, working memory, long-term vector store) and answer with explicit citations to source documents and character ranges. Use when asked to: find information, look up, search the workspace, retrieve, locate documents, what does the workspace say about, cite sources, where in the docs, ground the answer.

A bundled sample skill that ships with the ragbot demo. Demonstrates the SKILL.md format and shows up in `ragbot skills list` when RAGBOT_DEMO=1. Replace with your own skills when building a real knowledge base; see the synthesis-skills repository for community examples.