Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add SapienXai/AgentOS --skill agentos-codexgit clone --depth 1 https://github.com/SapienXai/AgentOSWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/sapienxai/agentos/agentos-codex)<a href="https://agentmods.dev/skills/sapienxai/agentos/agentos-codex"><img src="https://agentmods.dev/badge/skills/sapienxai/agentos/agentos-codex/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/sapienxai/agentos/agentos-codex"><img src="https://agentmods.dev/badge/skills/sapienxai/agentos/agentos-codex.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00032 | $0.04935 |
| Opus 5.5 | $0.00013 | $0.01974 |
| Sonnet 5.5 | $0.00006 | $0.00987 |
| Haiku 4.5 | $0.00003 | $0.00494 |
Grade A, and why
agentos scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 378 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AgentOS Codex Skill
Use this skill for AgentOS changes. Keep it practical: inspect the current code first, use the existing OpenClaw boundary, and make the smallest change that improves real operator control.
Product North Star
AgentOS is the human operating layer above OpenClaw. OpenClaw remains the backend, orchestration, runtime, agent/session/task/model/device/integration layer.
AgentOS must not become a replacement orchestrator or an OpenClaw clone. It should expose, visualize, manage, and improve UX around OpenClaw-backed workspaces, agents, tasks, files, models, integrations, policies, approvals, cost visibility, and runtime state.
Success question: did this system take a real workload from the owner today?
Start Here
Before implementing, inspect the relevant local surfaces:
- Product and architecture:
README.md,docs/openclaw-sync-audit.md,docs/openclaw-gateway-first-migration.md - Current OpenClaw contract and certification:
docs/openclaw-2026.10.6-upgrade-plan.md(prior 2026.10.5 and historical 2026.9.7 / 2026.9.4 evidence remains in versioned audit documents) - Native workspace foundation:
docs/openclaw-native-workspace-foundation.md - OpenClaw boundary:
lib/openclaw/client/,lib/openclaw/adapter/,lib/openclaw/application/,lib/openclaw/domains/ - AgentOS state/contracts:
lib/agentos/,hooks/use-mission-control-data.ts,hooks/use-task-feed.ts - UI:
components/mission-control/,components/operations/,app/*/page.tsx - API routes:
app/api/ - Release/install:
packages/agentos/package.json,packages/agentos/scripts/check-release-consistency.mjs,install.sh,install.ps1,.github/workflows/release-agentos.yml - Tests:
tests/, especially OpenClaw boundary, gateway-first, release consistency, and CLI smoke tests
OpenClaw-First Rules
- Check whether OpenClaw already provides the capability through Gateway RPC, SDK/config, CLI, session/task APIs, model APIs, integration APIs, or device APIs.
- Prefer native Gateway/API integration over CLI.
- Use CLI only when no stable native Gateway/API path exists, or for existing install, recovery, gateway process control, and unsupported Gateway operations.
- Any CLI fallback must be explicit, observable, and surfaced through diagnostics, logs, UI state, or returned metadata with reason and recovery path.
- Do not duplicate OpenClaw concepts unless AgentOS needs a UI projection, cache, adapter, or workspace-local sidecar.
- Keep AgentOS-specific state separate from OpenClaw runtime state.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 6d ago Changed bd525b3a26f5
- 9d ago Changed 9a6118b71e13
- 28d ago Changed 4ff50ff8a96c
- 1mo ago Changed · +12 lines f6025679f81a
- 1mo ago Changed 0cf094d48d57
- 1mo ago Changed · +1 lines 424a69538d9c
- 1mo ago Changed 779f5d09df87
- 1mo ago Changed 00c68db9976f
- 1mo ago First seen · 365 lines · 32 tokens per session scan A 52b862cd0bd1
agentos is a skill published in the GitHub repository SapienXai/AgentOS (129 stars, last pushed 4d ago), licensed MIT. It adds 32 tokens to every session and 4,935 once invoked, about $0.0001 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-05.
Other skills, from other repositories
pipeline-conductor
Operating procedure for the kirocrew-pipeline-conductor agent. Use when a pipeline conductor session is seeded, inspected or debugged. You MANAGE one issue/PR pipeline on one repo: two columns per item (session working? item solved?), send intent, never re-derive a red yourself.
babysit
Use when a user asks to babysit, monitor, keep checking, keep an eye on, or report when a pull request, CI run, ticket, deployment, or other changing target reaches an outcome.
feature-request
Conversational workflow for gathering user feedback and filing GitHub Issues on the Kiro Crew repository. Load when the user clicks "Request a Feature", wants to report a bug, or suggest an improvement.
goal-loop
Bootstrap a goal-driven self-improving AutoNudge loop from a goal plus an anchor directory, then run the board autonomously until the Definition of Done is met.
team-lead
Run a goal as a team when you can also do the work yourself: decide what you keep, dispatch the rest, and keep the status board honest. The delta over goal-conductor, which carries the dispatch, patrol and acceptance procedure this one points at. Load it when you are handed a goal, not a task.
goal-ledger-conductor
Deprecated alias of goal-conductor, removed next release. The work-ledger conducting procedure now lives in goal-conductor; read that skill instead.