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 ninthwave-io/ninthwave --skill friction-triagegit clone --depth 1 https://github.com/ninthwave-io/ninthwaveWrote 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/ninthwave-io/ninthwave/friction-triage)<a href="https://agentmods.dev/skills/ninthwave-io/ninthwave/friction-triage"><img src="https://agentmods.dev/badge/skills/ninthwave-io/ninthwave/friction-triage/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/ninthwave-io/ninthwave/friction-triage"><img src="https://agentmods.dev/badge/skills/ninthwave-io/ninthwave/friction-triage.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 | $0.00123 | $0.03746 |
| Opus 5 | $0.00062 | $0.01873 |
| Sonnet 5 | $0.00025 | $0.00749 |
| Haiku 4.5 | $0.00012 | $0.00375 |
Grade A, and why
friction-triage 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 8d 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 — 274 lines — stays where its author put it; the contents beside it link to each section on GitHub.
What this skill is
This is the canonical way to process friction logs in this repo. It is a synchronous, human-in-the-loop session: you read each log, the user decides what to do, you execute, you delete. There is no async batch / review-PR fallback. The whole point is that the human stays in the loop and the inbox actually gets to zero in one sitting.
Interactive Questions (CRITICAL)
This skill is highly interactive. You MUST use your interactive question tool to ask the user questions. Do NOT print a question as text and wait for a response.
Tool names by platform: AskUserQuestion (Claude Code), question (OpenCode), request_user_input (Codex), ask_user (Copilot CLI, Gemini CLI). Use whichever is available in your environment.
Every question must follow this structure:
- Re-ground: State the project, the current branch, what phase you are in, and "log M of N". Assume the user has not looked at this window in 20 minutes.
- Explain simply: Paraphrase the friction in plain English. Say what hurt, not just what the log file is named.
- Recommend: State which option you would pick and why, in one line.
- Options: Lettered options A) through F). When an option involves effort, indicate the scale.
Hard rules
- Never group questions across logs. Always one question per log. But DO batch the slow parts: investigation runs in parallel up front (Phase 2), execution runs after all decisions are collected (Phase 4). The human's interactive window in Phase 3 must be as tight as possible: no nested confirmations, no waiting for subagents, no follow-up "are you sure" prompts.
- Never commit. The skill stages edits and writes new files. The human commits at the end if they want to.
- Work items must follow
.ninthwave/work-item-format.md. Read it before writing any new work item file. - Generate lineage tokens with
nw lineage-tokenwhen creating a work item, the same way/decomposedoes. - Duplicate detection is best-effort. If you are not certain two logs are duplicates, surface it as an option and let the human decide. Do not auto-merge.
- "Drop" is a valid answer. Friction that is no longer relevant is exactly what this flow exists to clear. Do not argue with the user when they pick D.
- ASCII only. No em dashes, no smart quotes, no ellipsis characters. Use
--, straight quotes, and.... The repo enforces ano-em-dashlint rule on.mdfiles.
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.
- 8d ago First seen · 274 lines · 123 tokens per session scan A 32910468b6b7
friction-triage is a skill published in the GitHub repository ninthwave-io/ninthwave (8 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 123 tokens to every session and 3,746 once invoked, about $0.0006 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
armada-gate
Run an evidence-gate checklist for each success criterion in the current phase. Use when checking if a phase is complete before marking it passed.
inbound-triage
Triage raw human-channel input pasted directly into the session (chat exports, email threads, PR comments, meeting notes) — or dropped as files in a project inbox — into typed, deduplicated, decision-ready items with numbered recommendations. Strictly selection-gated — nothing dispatches without an explicit operator…
declare
Analyzes task plan and declares files that will be created or modified, producing declared-files.json.
execute
Executes the implementation plan for a Formic task.
sdd-tasks
Break an SDD change into implementation tasks. Trigger: orchestrator launches task planning for a change.
batch-all-issues
Resolve every open issue one at a time: fact-check each with web research, close the ones that need no action, and run the goal-pr skill to fix, review, and merge the ones that do — repeating until no actionable issues remain.