sw-explain

A plain-language guide to Session Watcher metrics that describe restart timing, cost, and the lowest point of a changing cost curve. It explains terms such as movable fraction, timing penalty, and sweet spot.

In plain words
What is it for?
Use it when interpreting metrics such as br, mf, pp, u, wall, sweet, or valley, and when deciding whether a restart is urgent.
Why use it?
It makes abbreviated dashboard labels understandable and puts timing costs in context, including whether a session is near its lowest-cost region.

Skill for Claude CodeCodex

Part of the session-watcher plugin — 3 skills, 1 hook shipped together

Install

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.

agentmods
npx agentmods add skills/nomadop/session-watcher/sw-explain
Any agent
npx skills add nomadop/session-watcher --skill sw-explain
Clone the repo
git clone --depth 1 https://github.com/nomadop/session-watcher

Made for: Claude Code, Codex.

Or install session-watcher, the plugin that ships this one along with the rest of its 3 skills, 1 hook.

Per session 54 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 395 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00054 $0.00395
Opus 5 $0.00027 $0.00198
Sonnet 5 $0.00011 $0.00079
Haiku 4.5 $0.00005 $0.00040

Measured 2d ago against content hash 7feba52421aa, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

sw-explain 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.

skills/sw-explain/SKILL.md · 24 lines

What it actually says

Explain a Session Watcher metric

When the user asks about a metric:

  1. Call watcher_status to confirm the dashboard is running and get its URL (metric VALUES live on the dashboard the user can see — do not expect raw numbers from MCP).
  2. Explain the requested metric in plain language.
  3. Contextualize against the flat-valley / AM-GM bound where relevant.

Metric glossary

  • br (bill regret) — how much extra you pay vs optimal restart timing, as a fraction. 0.10 = amber, 0.25 = red. br = mf × pp.
  • mf (movable fraction) — the share of cost that timing can affect. Capped at √2−1 ≈ 41.4% by AM-GM — even worst-case timing can't cost more than that.
  • pp — the timing-penalty shape term (u−1)²/(2u); minimized at u=1 (the sweet spot).
  • u — normalized position (x−1)/dhat. u<1 = left arm (cost still falling, no action). u=1 = sweet. u>1 = past sweet.
  • wall — where continuing costs more per turn than a full restart.
  • sweet / valleyxSweet = 1 + dhat, the cost-curve minimum; the valley around it is flat (small timing penalty).

Frame reassuringly: e.g. "br=18% sounds high, but the ceiling is ~41.4%, so you're inside the flat valley — no urgency."

Changes

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.

  1. 2d ago First seen · 24 lines · 54 tokens per session scan A 7feba52421aa

Subscribe to this mod's changes

sw-explain is a skill published in the GitHub repository nomadop/session-watcher (3 stars, last pushed 1mo ago), licensed MIT. It adds 54 tokens to every session and 395 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 2026-08-31.

Related

Other skills, from other repositories

dvalin-security-scan

Scan code for injection, hardcoded secrets, XSS, dynamic code execution, and unsafe shell use. Use after writing or modifying code that handles user input, builds queries or commands, touches authentication, or adds a dependency — and whenever the user asks for a security check, audit, or review. Runs locally with no…

arthurpanhku/dvalincode · 77 tokens

coding-agents-farm

To orchestrate parallel coding-agent farms (Claude, Codex, Copilot, Gemini, etc.) on isolated git worktrees.

griddynamics/rosetta · 32 tokens

c-03-repo-bootstrap

Bootstrap onboarding for undocumented repos or existing memory slices. Builds root overviews, route-local overview pillars, evidence packs, file cards, onboarding waves, deleted-slice cleanup, curator reviews, and handoff while keeping the orchestrator thin.

Foxfire1st/agents-remember · 54 tokens

c-12-closeout

Close out approved Agents Remember edits by preserving approval authority, mandatory strict code quality before code commit, missing-onboarding checks, external-memory refresh, memory quality, ledger alignment, and no automatic push.

Foxfire1st/agents-remember · 45 tokens

l-01-agent-lifecycles

The agent lifecycles: one lifecycle per agent type, under one roof. Routes every session by exactly three conditions (spawn-role env -> fresh role brief -> otherwise free-chat launcher), carries the minimal lifecycle frame (the six lifecycle signals every session shares), and houses the self-contained per-role…

Foxfire1st/agents-remember · 135 tokens

c-09-git-worktree-manager

Create, attach to, report on, integrate, finalize, and clean up Agents Remember worktree-backed tasks while preserving delegated/human approval gates and external-memory compatibility.

Foxfire1st/agents-remember · 42 tokens