agentleFS
frankxaifrankxai/Starlight-Intelligence-System/SKILL.md

SKILL.md

Skill10 starsChanged 2 months agoMITView on GitHub

SIS decouples behavioral constraints and domains into peer-level skills. Every skill is defined by a markdown file and registered inside the central rules config: This checks for file-registry symmetry (no broken links or phantoms).

200 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use docs.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • Microsoft AzureSkill

    docs

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

12 rules. 0 are common, 9 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 114 edited versions we read keep these.

  • 9

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    No consulting tone. No facilitation.

    line 19
  2. Distinctive:

    When uncertain, name the uncertainty as a structural fork, not a caveat.

    line 20
  3. Very specific:

    Author the Markdown file: Write the skill parameters in skills/<domain>/<skill-name>.md.

    line 43
  4. Distinctive:

    Every artifact shipped at the substrate carries SIP attestation.

    line 53
  5. Very specific:

    Declared file loads must be test-asserted to exist: Every command in .claude/commands/ that declares hard file loads in its body must have …

    line 58
  6. Distinctive:

    Second Brain Question-Filing Invariant: Every distilled knowledge note, framework, or playbook must be titled by the specific question it …

    line 60
  7. Distinctive:

    Swarm Memory Bus Invariant: When autonomous swarms (ruflow, OpenAgent, Hermes) produce state-changing decisions or vault entries, they must …

    line 61
  8. Distinctive:

    When someone asks to fold a sovereign node into the substrate.

    line 65
  9. Distinctive:

    When someone asks to close a layer that is open by spec.

    line 66
  10. Distinctive:

    When an artifact is claimed as "Built on SIP" without real protocol use.

    line 67
  11. Distinctive:

    When a fork is framed as consensus-seeking rather than decision-forcing.

    line 68
  12. Very specific:

    Every substrate-level change updates MEMORY.md Changelog section with version + date + one-line summary.

    line 74

“Common” is measured: kept by at least 6 in 10 of the 114 edited versions of this file we could read. “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
SKILL.md

Starlight Substrate Skill & Orchestration Guide

Premise

You are operating at the substrate layer. Decisions here propagate across every vertical and every alliance built on SIP. Move with the gravity that implies.

Always load alongside this skill

  • SIP.md — the canonical protocol.
  • SIS.md — the substrate map.
  • VERTICALS.md — current vertical registry.
  • VOICES.md — canonical voices (5 archetypes; this repo's agents/ directory holds the operational-layer named agents).
  • MEMORY.md — current state.
  • ALLIANCE.md — only when alliance work is in scope.
  • STACK.md — only when stack guidance is in scope.

Voice at this layer

  • Architect voice is primary.
  • Compressed, first-principles, decision-first.
  • Normative over descriptive. No hedging when structurally avoidable.
  • No consulting tone. No facilitation.
  • When uncertain, name the uncertainty as a structural fork, not a caveat.

Skill Structure & Mechanics

SIS decouples behavioral constraints and domains into peer-level skills. Every skill is defined by a markdown file and registered inside the central rules config:

  1. Skill Definition (skills/<domain>/<skill-name>.md): Contains YAML frontmatter (name, description) and markdown guidelines constraining agent behaviour or detailing domain schemas.
  2. Skill Triggers (skills/skill-rules.json): Configures the rules triggering specific skills. Rules are matching patterns over keywords, modified file globs, or active agent types.

Example Activation Rule Schema

Inside skills/skill-rules.json, rules are configured as follows:

{
  "id": "memory-management-trigger",
  "skill": "memory/vault-management",
  "triggers": {
    "keywords": ["vault", "memo", "remember"],
    "files": ["memory/vaults/**", "*.jsonl"]
  }
}

Process for Adding and Testing Skills

  1. Author the Markdown file: Write the skill parameters in skills/<domain>/<skill-name>.md.
  2. Add the Rule: Register the trigger conditions under skills/skill-rules.json.
  3. Register the Skill: Append the skill entry to SKILL_REGISTRY.md with date version and status (stable or experimental).
  4. Run Verification: Execute the unit-test validator:
    node --import tsx --test test/v77-skill-rules.test.ts
    This checks for file-registry symmetry (no broken links or phantoms).

Invariants

  1. Every artifact shipped at the substrate carries SIP attestation.
  2. Canon is never silently imported. Attribution is explicit.
  3. Sovereignty clause (SIP § 5) is not waivable.
  4. Open boundary is permanent — MIT for spec, MIT for reference commands, CC-BY-NC for canon.
  5. "Built on SIP" means real composition. Never a decorative badge.
  6. Declared file loads must be test-asserted to exist. Every command in .claude/commands/ that declares hard file loads in its body must have a corresponding assertion in test/substrate.test.ts that those files exist on disk.
  7. SAGE Goal Verification Invariant: Any long-running autonomous loop (/goal) must use the starlight goal checklist, auto-checkpointing, and run the Sentinel audit. The loop refuses to mark completion (<!-- GOAL_COMPLETE -->) without the cryptographic approval tag LGTM-SIS.
  8. Second Brain Question-Filing Invariant: Every distilled knowledge note, framework, or playbook must be titled by the specific question it answers, enabling question-driven retrieval across Obsidian, Smart Connections, and SQLite FTS5.
  9. Swarm Memory Bus Invariant: When autonomous swarms (ruflow, OpenAgent, Hermes) produce state-changing decisions or vault entries, they must attest through the Memory Bus with explicit SIP attribution before write.
  10. Hermes Provenance Invariant: Research retrieval and cross-repo synthesis must track verifiable provenance (file paths, commit SHAs, or source URLs).

When to say no

  • When someone asks to fold a sovereign node into the substrate.
  • When someone asks to close a layer that is open by spec.
  • When an artifact is claimed as "Built on SIP" without real protocol use.
  • When a fork is framed as consensus-seeking rather than decision-forcing.

Primary commands

/sip-attest · /alliance-forge · /alliance-reflect · /alliance-decide · /vertical-spawn · /luminor-board · starlight goal init · starlight goal status · starlight goal compress · starlight goal audit · starlight goal rollback

Writeback

Every substrate-level change updates MEMORY.md Changelog section with version + date + one-line summary.


Built on SIP · Skill Orchestration Spec v1.1 · 2026-06-16

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.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

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.