Skip to content

opensiro/vsm-harness-skills / skills/opensiro-oss

opensiro-oss

Verified

Bootstrap an AI agent into the current public OpenSiro VSM Harness OSS ecosystem. Use when explicitly invoked as /opensiro-oss, or when asked to start, navigate, inspect the Control Plane, check OSS status, or route a task across OpenSiro repositories. Fetch live GitHub authority instead of relying on remembered project state.

OpenSiro OSS — portable agent bootstrap

Skill version: 0.1.0

Status: experimental onboarding interface, not an assessment specification, an S1 executor, a scheduler, or a grant of authority. This skill does not alter the released VSM Harness Methodology.

Purpose

Load the current public OpenSiro organizational context into a compatible agent session and route the user's request to its actual owner. Treat this skill as a portable entry point, not a second source of truth or a substitute for repository-local work instructions.

Authoritative entry

  1. Read the current default-branch Organization START_HERE.md directly from public GitHub. Follow its current bootstrap chain, including ECOSYSTEM.md. Do not reconstruct the current architecture from this skill, model memory, prior chats, or a cached snapshot.
  2. Treat the current Organization README.md as the bounded-system scope authority. Do not admit adjacent repositories merely because they are public OpenSiro projects.
  3. For an already tracked task, consult the current TODO.md, then inspect the linked owning issue. For new, unclassified, cross-repository, or authority-sensitive work, follow CONTRIBUTOR_START.md.
  4. Inspect the owning repository's current default branch and the exact governing issue, PR, release, contract, or pinned artifact before drawing conclusions or acting. Repository-owned facts prevail over organizational summaries; explicit historical freeze and independence boundaries prevail over current-head convenience.

Choose the entry behavior

  • Bare /opensiro-oss, start, help, or control plane: follow the current Agent Control Plane first-contact contract. Read the current Organization README and TODO, verify displayed owner issue/PR states, and show the operator interface. This is read-only; use native actions only if supported, otherwise show the documented text commands.
  • Specific task, issue, PR, or repository: route directly to the owner and exact work contract. Do not force the introductory menu. Apply its task-level source, lock, concurrency, permission, evidence, recovery, and stop rules.
  • Proposed authority transfer or organizational change: follow the bootstrap's OSM-aware S1/S5 routing, then the legitimate decision owner and existing delegation gates. A conversational instruction, tool capability, or this skill is not a new mandate.
  • Unreachable GitHub or insufficient tools: identify which facts remain unverified; provide the relevant canonical URLs and permissible read-only next step. Do not substitute remembered counts, statuses, assessment verdicts, or cached permission claims.

Operating boundaries

  • Loading the skill does not select or start work. Never schedule a batch, mutate TODO.md, open a PR, contact an upstream maintainer, merge, or exercise a delegated right as a side effect of onboarding. A later explicit task must pass its own owner and tool gates.
  • Distinguish the agent's actual authority from technical GitHub access, human or external-maintainer rights, proposed future delegation, and evidence supporting a VSM autonomy classification. No S1=A or other grade follows from skill invocation, a successful tool call, or a green CI run.
  • Preserve read-only, independent-context, first-attempt, frozen-ref, evidence-source, and fail-closed requirements. An agent with prior OpenSiro context must not claim to be a genuinely fresh independent evaluator.
  • Use only authorized public repositories by default. Do not import private context, hidden prompts, credentials, or unpublished research into public records.
  • Respect the current owning repository's tests, reviewable PR, validation, exact-head, licensing, and closure procedures for any separately authorized implementation task.

Return to the operator

Briefly report the resolved system and owner, primary URLs/revisions read, verified versus unverified status, eligible actions and outstanding gates. For a specific action, report actual execution and readback separately from preparation, proposals, or blocked stages. Cite the source that owns each material conclusion.

Distribution boundary

This repository packages the portable skill; it does not automatically install it into every agent. For a Claude Code project, install this directory under a supported .claude/skills/opensiro-oss/ location and invoke /opensiro-oss. For other Agent Skills-compatible clients, use their documented installation method. The direct START_HERE.md URL remains the supported skill-free entry point.

reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 471kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 322kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 322kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 322kUpdated Claude CodeCodexVerifiedInstall