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 baek-labs/hames --skill setupgit clone --depth 1 https://github.com/baek-labs/hamesWrote 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/baek-labs/hames/setup)<a href="https://agentmods.dev/skills/baek-labs/hames/setup"><img src="https://agentmods.dev/badge/skills/baek-labs/hames/setup/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/baek-labs/hames/setup"><img src="https://agentmods.dev/badge/skills/baek-labs/hames/setup.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 | $0.00018 | $0.00695 |
| Opus 5.5 | $0.00007 | $0.00278 |
| Sonnet 5.5 | $0.00004 | $0.00139 |
| Haiku 4.5 | $0.00002 | $0.00069 |
Grade A, and why
setup 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 28d 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.
What it actually says
Setup
Use the user's language and ask one decision at a time. On first use, ask what work the user wants to do, which root to manage, workspace names/paths/purposes, and subfolder roles. Offer editable examples, never fixed personal categories. Existing folders may be registered in place. Reuse choices already known; an uncertain answer invites a recommendation, not an invented decision.
Turn the answers into workspace definitions: id, root-relative path, name, purpose, folders with workspace-relative path and purpose, rules with id/kind/path/value, exclude, context, and protect. Rule kinds are extension (list of extensions), name (regular expression), required (required text), and semantic (a judgment instruction). Explain concrete checks versus meaning-based judgments. Shared document registrations have root-relative docs/ path, purpose, scope (common or workspace id), and optional dependencies.
Use bundled runtime/setup.js plan --root <root> --input <decisions.json>. The input accepts projectName, workspaces, documents, exclude and contractTracking. Do not require users to write JSON; stage tool input outside their workspace until approval (host temporary storage), without secrets. The preview is read-only. Show the proposed tree, roles, file rules, exclusions, exact existing-file edits and initial indexes. Non-Git roots default to untracked contracts without initializing Git. A Git root with an unresolved tracking choice gets one question.
After the user approves that exact preview, run runtime/setup.js apply --root <root> --input <decisions.json> --approved --plan-hash <preview hash>. This is the only setup approval; do not ask twice. A changed preview is not the approved preview. Re-read the applied configuration and indexes and report any unresolved classification. Do not initialize this plugin's development repository as a user workspace.
For reconfiguration, load current choices, change only what the user requests, and show a new preview. Repeated valid setup is read-only. Changing a workspace path never moves the user's existing files. Missing registered folders and damaged settings need an explicit recovery choice; do not silently recreate them. Existing entry files, manual index text, shared docs and archived contracts are preserved. Active old-format contracts block upgrade until completed or explicitly handed off.
runtime/setup.js recover provides a read-only recovery preview. Only after the user chooses that recovery, pass --approved --recovery-hash <hash>. Never overwrite newer user edits while rolling back.
Recognized old public Hames folders use the existing manifest-based transition preview. Keep modified, unknown, protected and submodule files. Only exact unchanged old system files listed in the approved preview may be removed. Never expose protected content or hashes. Do not alter Git history, index, branches or remotes.
Hames already includes adapted DryForge workflows; do not install another DryForge plugin. Long-lived operating documents belong to shared root docs/, not each workspace. Do not generate a standard empty document set. Use /index for a read-only file audit, /doctor for configuration and hook diagnosis, and /ready then /go for bounded task execution. Ordinary file work needs no Git or task contract.
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.
- 28d ago Changed · -2 lines dab12d72f2c8
- 1mo ago Changed · +2 lines ad2fe483cf56
- 1mo ago First seen · 23 lines · 18 tokens per session scan A 83b4dd6dcff6
setup is a skill published in the GitHub repository baek-labs/hames (5 stars, last pushed 28d ago), licensed MIT. It adds 18 tokens to every session and 695 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 2026-09-04.
Other skills, from other repositories
b3os-team-inbox
A team message-bus client for sending, receiving, and recalling messages between AI teammates. It also supports replying to a specific message thread and sending final reports to the team lead.
b3os-task-loop
A task-tracking work loop that keeps individual, scheduled, and waiting tasks moving until they are completed, blocked, or awaiting approval.
b3os-sf
A software-development method for turning a growing list of team requests into planned, reviewed, tested, documented, and released work. It describes a project structure with a repository, design notes, task tracking, continuous integration, and agent collaboration.
b3os-task-mgmt
A deprecated compatibility stub that points users to b3os-task-loop, the newer source for task boards, execution mode, handoffs, continuation checks, review waiting, and scheduled work.
project-flow
A workflow guide for AI agents working in a project with defined roles, file permissions, and handoff records.
codex-harness-setup
Configure the shared tracker and explanation profile before generating a Codex harness, or update those project preferences. Reuses to-spec and to-issues .scratch/.tracker; writes the chosen audience profile and harness policy pointers into AGENTS.md. Does not execute work or publish issues merely because setup was…