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 musterhq/muster --skill muster-frappe-workflowsgit clone --depth 1 https://github.com/musterhq/musterWrote 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/musterhq/muster/muster-frappe-workflows)<a href="https://agentmods.dev/skills/musterhq/muster/muster-frappe-workflows"><img src="https://agentmods.dev/badge/skills/musterhq/muster/muster-frappe-workflows/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/musterhq/muster/muster-frappe-workflows"><img src="https://agentmods.dev/badge/skills/musterhq/muster/muster-frappe-workflows.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 | $0.00065 | $0.00717 |
| Opus 5 | $0.00032 | $0.00358 |
| Sonnet 5 | $0.00013 | $0.00143 |
| Haiku 4.5 | $0.00006 | $0.00072 |
Grade A, and why
muster-frappe-workflows 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 10d 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 — 39 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Muster Frappe Workflows
Build a universal workflow runtime, not a sequence of API calls. A workflow is a versioned graph of goals, tasks, agents, tools, policies, approvals, events, artifacts, and compensation.
Canonical Model
Every definition has versioned inputs, outputs, nodes, edges, policy bindings, budgets, approval rules, and compatibility metadata. Every run has an immutable event stream and a materialized current state.
Node kinds include plan, agent, subworkflow, Frappe command, deterministic transform, condition, parallel map, human approval, wait/event, artifact, verification, and compensation. Agent nodes may spawn bounded children; inheritance of tenant, identity, permissions, budget, and tool allowlists must be explicit.
Execution is at-least-once. Every attempt uses a stable operation idempotency key and a lease fencing token; a stale worker cannot commit events or effects. Events receive a monotonic sequence from the authoritative store, consumers resume from cursors, and duplicate delivery must be harmless. The default safety envelope is depth 3, 8 children per node, 32 active nodes per mission, and no unbounded tokens/time/cost; policies may lower or explicitly raise these limits.
Raw graph cycles are invalid. Repetition uses a bounded loop node with an iteration cap, progress predicate, budget, and cancellation checkpoint. Cancellation prevents new effects, propagates to descendants, and waits for in-flight operations to reach a declared safe point. Compensation is a separately recorded best-effort workflow; a failed compensation leaves the mission needs intervention, never falsely rolled back.
Workflow
- Audit
.workflows, Claude Code workflow conventions, Muster run/router/roster code, and existing Frappe capability packs before adding a format. - Define a portable JSON schema and a JS/TypeScript authoring API that compile to the same intermediate representation.
- Validate graphs for raw cycles, unreachable nodes, invalid schemas, excessive depth/fan-out, unbounded loops, incompatible permissions, missing compensation, and unsafe triggers.
- Persist definitions separately from runs. Pin each run to an immutable definition version.
- Execute through durable queues with leases, heartbeats, idempotency keys, bounded retries, timeouts, cancellation propagation, and resume.
- Stream sanitized events to Frappe while retaining full authorized evidence server-side.
- Test nested failure, partial success, duplicated delivery, worker death, approval expiry, concurrent edits, and compensation.
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 10d ago First seen · 39 lines · 65 tokens per session scan A 7aacac48df94
muster-frappe-workflows is a skill published in the GitHub repository musterhq/muster (11 stars, last pushed 5d ago), licensed MIT. It adds 65 tokens to every session and 717 once invoked, about $0.0003 per session on Opus 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 2026-08-30.
Other skills, from other repositories
kungfu-agent-onboarding
Use when a user asks to understand, start, inspect, extend, or safely operate installed Kungfu; verify the installed pack, select one intent route, personalize the explanation, and propose one smallest safe next action.
python-run
Run and debug Python scripts in the project. Use when the user says "run python", "execute this script", "debug this py file", or wants to run/modify a .py file. Handles dependency checks, linting, execution, and error analysis.
efficient-execution
Reduce token waste and task completion time by choosing the simplest tool for each job.
task-planning
Match planning effort to task complexity. Check existing skills before inventing a new approach.
learning-capture
Log corrections, errors, and discoveries to orbital/LESSONS.md so future sessions avoid repeating mistakes.
process-capture
Capture successful multi-step workflows as reusable skills so future sessions can replay them.