agentleFS
aidvggaidvgg/operator-os/.github/copilot-instructions.md

operator-os

Copilot instructions4 starsChanged 2 months agoMITView on GitHub

Platform surface: GitHub Copilot. Read CLAUDE.md in full before acting. It is the canonical shared operating contract for every model, not a Claude-only file. It keeps the filename CLAUDE.md so the Claude Code loader still finds it without configuration. Then follow its context routing, precedence, write, safety, and verification rules exactly. The aliases and fallbacks below are generated from ai/AGENT_ROUTES.json, their single owner. Canonical workflow: skills/daily-log/SKILL.md. Aliases: /plan, plan my day, plan the day, let's plan my day, let's plan…

We split this copilot-instructions.md into the rules it gives an agent

5 rules. About 0 are common, 4 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read the canonical daily-log skill before responding.

    line 22
  2. Distinctive:

    Read the live ledger, carry-over, owner files, working-tree evidence, and horizon output required by that skill before asking for generic …

    line 23
  3. Distinctive:

    Never invent tasks. The plan may contain only Sam-stated tasks, carried work, and standing declared contracts.

    line 25
  4. Distinctive:

    Every skill this repository ships, with its trigger and its canonical file:

    line 67
  5. Very specific:

    Do not edit this file directly.

    line 97

≈ “Common” is an estimate: at least 1 in 5 of the 3,669 Copilot instructions files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
copilot-instructions.md
<!-- GENERATED by scripts/sync-agent-adapters from ai/AGENT_ROUTES.json. DO NOT EDIT. -->

Northwind Labs agent entrypoint

Platform surface: GitHub Copilot.

Read CLAUDE.md in full before acting. It is the canonical shared operating contract for every model, not a Claude-only file. It keeps the filename CLAUDE.md so the Claude Code loader still finds it without configuration. Then follow its context routing, precedence, write, safety, and verification rules exactly.

Immediate workflow routing

The aliases and fallbacks below are generated from ai/AGENT_ROUTES.json, their single owner.

daily-log

Canonical workflow: skills/daily-log/SKILL.md.

/plan

Aliases: /plan, plan my day, plan the day, let's plan my day, let's plan the day, daily plan, daily planning.

First-turn contract for /plan only:

  • Read the canonical daily-log skill before responding.
  • Read the live ledger, carry-over, owner files, working-tree evidence, and horizon output required by that skill before asking for generic intake.
  • Surface known commitments and conflicts first, then ask only for missing day-specific tasks or hard stops.
  • Never invent tasks. The plan may contain only Sam-stated tasks, carried work, and standing declared contracts.

Capability fallbacks for /plan only:

  • Without calendar read access, ask Sam for fixed events after loading repository context, then continue planning.
  • Without calendar write access, continue planning and state that no calendar blocks were synced. Never invent event IDs.
  • Without repository write access, produce the exact plan draft and state that it was not logged.

/standup

Aliases: /standup, standup, daily standup.

Capability fallbacks for /standup only:

  • Without calendar read access, omit the optional next calendar item and continue the repository-backed standup.

/eod

Aliases: /eod, eod, log my day, end of day, close the day.

Capability fallbacks for /eod only:

  • Without repository write access, produce the exact end-of-day update and accountability diff, state that it was not logged, and do not claim files changed.

/weekly

Aliases: /weekly, weekly review, weekly retro.

Capability fallbacks for /weekly only:

  • Without repository write access, produce the exact weekly review draft, state that it was not logged, and do not claim files changed.

On a semantic match, read the canonical workflow before responding and execute only the matched mode. Do not substitute a generic intake checklist or borrow another mode's first-turn behavior.

Skill discovery

Skills hosts discover repository skills under .agents/skills/ (Codex, invoked as $<name> or through /skills) and .claude/skills/ (Claude Code). Each of those files is a generated discovery adapter that names its canonical file under skills/, and the canonical file owns the procedure. Read the canonical file in full before running a skill; never execute an adapter's summary as if it were the skill. Claude Code additionally reaches some skills as slash commands under .claude/commands/; those command files are thin pointers to the same canonical files.

Every skill this repository ships, with its trigger and its canonical file:

  • analytics-tracking (vendored): Set up, improve, or audit analytics and measurement: GA4, Google Tag Manager, conversion and event tracking, UTM parameters, tracking plans. Canonical file: skills/analytics-tracking/SKILL.md.
  • app-store-screenshots (vendored): Build App Store screenshot pages and programmatic marketing-screenshot generators for an iOS release. Canonical file: skills/app-store-screenshots/SKILL.md.
  • business-profile-creator (first-party): Update the operator's business profile JSON by read-and-merge. Trigger on update the business profile, refresh the profile JSON. Reads the existing profile and memory.md before asking anything. Canonical file: skills/business-profile-creator/SKILL.md.
  • copy-editing (vendored): Systematic multi-pass edit of existing marketing copy. Trigger on edit this copy, review my copy, proofread, polish this, copy sweep. Canonical file: skills/copy-editing/SKILL.md.
  • copywriting (vendored): Write or rewrite marketing copy for a homepage, landing, pricing, feature, about, or product page. Trigger on write copy for, rewrite this page, headline help. Canonical file: skills/copywriting/SKILL.md.
  • hard-task (first-party): Tier-2 protocol for high-stakes work no other skill covers: legal, tax, banking, entity, relocation, canonical knowledge rewrites, repo reorganization, irreversible client-facing moves. Canonical file: skills/hard-task/SKILL.md.
  • harvest (first-party): End-of-session content harvest. Trigger on harvest this session, pull content seeds, any content in this. Writes 1-3 receipt-backed X-voice seeds to outputs and never posts. Canonical file: skills/harvest/SKILL.md.
  • invoice-creator (first-party): Issue or reissue a branded invoice. Trigger on invoice this client, bill for agreed work, send an invoice. Takes the number from the ledger and the wire block from the payment file, never from recall. Canonical file: skills/invoice-creator/SKILL.md.
  • landing-page (vendored): Design or rewrite a high-converting single-offer landing page: structure, layout patterns, conversion strategy, copy, SEO, common pitfalls. Canonical file: skills/landing-page/SKILL.md.
  • make-interfaces-feel-better (vendored): Design-engineering details that make an interface feel polished: animation, hover states, shadows, borders, typography, optical alignment, micro-interactions. Canonical file: skills/make-interfaces-feel-better/SKILL.md.
  • ogilvy (vendored): David Ogilvy's advertising principles for copy that sells: positioning, headlines, promises, brand voice, long-form copy, visual logic. Canonical file: skills/ogilvy/SKILL.md.
  • onboard (first-party): Set up Operator OS for a new operator: first-run steps, plain-language interview, rebrand, demo wipe, verification, first snapshot. Trigger on set up Operator OS for me, make this mine, onboard me, first run after cloning. Canonical file: skills/onboard/SKILL.md.
  • operational-enterprise-ai (vendored): Enterprise AI, automation, security, and operations product pages: system boundaries, approvals, auditability, exceptions, rollback, security proof. Canonical file: skills/operational-enterprise-ai/SKILL.md.
  • page-cro (vendored): Conversion-rate optimization for any marketing page. Trigger on CRO, this page is not converting, improve conversions, why is this page not working. Canonical file: skills/page-cro/SKILL.md.
  • pricing-page (vendored): Design or rewrite a high-converting SaaS pricing page: plan design, structure, copy, FAQs, layout patterns, experiments. Canonical file: skills/pricing-page/SKILL.md.
  • product-proof-saas (vendored): SaaS and AI product landing pages where a real workflow, interface, or deterministic demo is the central proof rather than a broad claim. Canonical file: skills/product-proof-saas/SKILL.md.
  • proposal-creator (first-party): Generate a branded proposal, quote, SOW, or service agreement as print HTML rendered to PDF. Trigger on create a proposal, write a quote, SOW for a project, proposal for a client. Canonical file: skills/proposal-creator/SKILL.md.
  • prospect-brief (first-party): Turn a profile URL or a name into a decision-ready outbound prospect brief with a qualification-gate verdict, a pain-bank match, and opener prep. Trigger on brief this prospect, triage these accepts, screen this batch. Canonical file: skills/prospect-brief/SKILL.md.
  • research-doctrine (first-party): Epistemic assurance for research, synthesis, validation, due diligence, and decisions. Trigger when external evidence, current facts, multiple sources, disputed claims, recommendations, or silent omissions could matter. Canonical file: skills/research-doctrine/SKILL.md.
  • schema-markup (vendored): Structured data and JSON-LD schema markup for search engines and AI answer surfaces, including validation and common markup mistakes. Canonical file: skills/schema-markup/SKILL.md.
  • seo-audit (vendored): Audit a site for technical SEO, on-page, and content issues, then rank the fixes by impact. Trigger on SEO audit, why are we not ranking, check our SEO. Canonical file: skills/seo-audit/SKILL.md.
  • stop-slop (vendored): Strip AI-slop tells out of drafted copy. Supplement to the binding copy rules in knowledge/voice/copy-rules.md, never a replacement for them. Canonical file: skills/stop-slop/SKILL.md.
  • tailwind-v4 (vendored): Tailwind CSS v4 conventions, configuration, and migration for client-build and site work. Canonical file: skills/tailwind-v4/SKILL.md.
  • visual-style-presets (vendored): Named visual style presets for choosing and committing to a coherent look on a page or app build. Canonical file: skills/visual-style-presets/SKILL.md.
  • web-design-guidelines (vendored): General web design and build guidelines for pages and interfaces: layout, spacing, hierarchy, responsive behavior, accessibility. Canonical file: skills/web-design-guidelines/SKILL.md.

Maintenance

Do not edit this file directly. Change the owning contract, workflow, or ai/AGENT_ROUTES.json, then run scripts/sync-agent-adapters --write. scripts/repo-doctor blocks stale generated surfaces.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

Reviews can't be read right now.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.