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 agentmods add skills/lengros/bennu/reframenpx skills add Lengros/bennu --skill reframegit clone --depth 1 https://github.com/Lengros/bennuWhat 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 | $0.00148 | $0.01974 |
| Opus 5 | $0.00074 | $0.00987 |
| Sonnet 5 | $0.00030 | $0.00395 |
| Haiku 4.5 | $0.00015 | $0.00197 |
Grade A, and why
reframe 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 2d 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 — 156 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/reframe — output → outcome problem framer
Produces one structured artifact that converts a request into a verifiable problem statement. It does not do the work. It specifies what the work must achieve so it can be checked, delegated, or argued about before a single hour is spent executing.
The engine, repeated at every stage: the request is an output (a chosen action). The task is the outcome behind it (an observable change in the world). Never execute the output until the outcome is named and made falsifiable.
Argument (optional): the raw request. If absent, use the most recent user ask in context.
How to run
Work the stages in order. Each feeds the next. Output the YAML-style artifact in §Schema. Most stages are 1–3 sentences — resist padding; one idea per line. One stage (Validate direction) is an interactive checkpoint — it can hand control back to the user mid-pass.
-
Capture — record the request verbatim in
original_request. Do not improve it. -
Root cause (5 Whys) —
root_cause. Name what the request is ("congratulate" is an output — a chosen action), then ask why until you hit the driver — the change the user actually wants in the world. Stop when the next "why" would leave this person's actual situation. Tag the driver:[Observed: …]if the conversation supplies it,[ASSUMPTION]if you inferred it. An untagged root cause is a defect. -
Validate direction (interactive, fires on a fork) — before spending criteria on the root cause, enumerate the candidate readings of it. Then:
- If two or more readings lead to materially different acceptance criteria → this is a fork. Stop and ask the user. Surface your lead hypothesis first (marked), 1–3 genuine alternatives, and "or your own". Use the platform's multiple-choice question prompt so the user picks rather than free-types. Iterate up to ~2 rounds — each round narrows from the user's answer — until they confirm or correct the direction. Converge in one round when the lead hypothesis is confirmed.
- If only one reading survives → proceed, but record the alternatives you considered and
rejected so the user can object, and keep the
[ASSUMPTION]tag. Bias: when unsure whether a fork is material, treat it as material and ask. Divining the wrong root cause produces confident, precisely-wrong ACs — the exact failure this skill exists to prevent. Skip this stage only when the user has already pinned the outcome explicitly in the conversation. Record the settled direction invalidated.
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.
- 2d ago First seen · 156 lines · 0 tokens per session scan A ead0eec69912
reframe is a skill published in the GitHub repository Lengros/bennu (2 stars, last pushed 1mo ago), licensed MIT. It adds 148 tokens to every session and 1,974 once invoked, about $0.0007 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 2026-08-31.
Other skills, from other repositories
feishu
Work with Feishu or Lark bots, docs, sheets, bitables, approval flows, and OpenAPI/MCP setup without hardcoding credentials.
interview
Ask one useful structured question at a time only when material product/implementation choices are genuinely missing; remember answers and produce a brief/spec. Discoverable facts should be investigated instead of asked.
local-frontend-check
Smoke-test or verify UI behaviour on the local Jarvis Registry frontend running at http://localhost/gateway. Use for manual regression checks, bug-fix verification, and end-to-end confirmation of specific flows without running the automated test suite.
ppt-review
Read this skill only after current officereview structural output and relevant rendered images exist. Do not use source code, a tool success flag, or the first-slide preview as a substitute for deck evidence.
material-organizer
Organize only the materials the user has already provided or explicitly pointed to. The job is to make messy inputs easier to scan, reuse, archive, or continue from later while preserving source traceability.
helmor-debug-operate
Operate, reproduce, and debug a running local Helmor desktop development build through the Tauri MCP bridge. Use when the user asks to use Tauri MCP, towery MCP, the local dev build, the Tauri webview, visual end-to-end validation, UI automation, screenshots, DOM/accessibility snapshots, IPC or log tracing…