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 mostlyharmless-ai/watercooler --skill update-agent-contextgit clone --depth 1 https://github.com/mostlyharmless-ai/watercoolerWrote 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/mostlyharmless-ai/watercooler/update-agent-context)<a href="https://agentmods.dev/skills/mostlyharmless-ai/watercooler/update-agent-context"><img src="https://agentmods.dev/badge/skills/mostlyharmless-ai/watercooler/update-agent-context/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/mostlyharmless-ai/watercooler/update-agent-context"><img src="https://agentmods.dev/badge/skills/mostlyharmless-ai/watercooler/update-agent-context.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 | $0.00203 | $0.13698 |
| Opus 5.5 | $0.00081 | $0.05479 |
| Sonnet 5.5 | $0.00041 | $0.02740 |
| Haiku 4.5 | $0.00020 | $0.01370 |
Grade A, and why
update-agent-context 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 1mo 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 — 1,098 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Update Agent Context
Overview
Keeps CLAUDE.md and AGENTS.md short, current, and auditable. CLAUDE.md is the
source of truth; AGENTS.md is derived from it. The skill edits the working tree and
shows a diff — it does not commit. Use /ship to commit after review.
Arguments: /update-agent-context [--phase1 | --phase2 | --phase3 | --all]
--all runs Phase 1, then Phase 2, then Phase 3 in sequence. Default when no
flag is given: ask the user which phase to run.
Phase 3 is read-only — it audits the skill surface and produces a punch list. It does not edit any file. Run it after any watercooler-cloud tool-surface consolidation, or whenever a skill call unexpectedly fails because its tool name no longer exists.
Phase 1 — Structural Refactor or From-Scratch Seed
Run to (re)build CLAUDE.md on the approved structure and derive a fresh
AGENTS.md. Phase 1 has two modes; Step 0 selects between them.
Step 0: Select mode
Check whether CLAUDE.md exists at the repo root.
- Refactor mode —
CLAUDE.mdexists. The current file is the primary content reservoir: carry its section content forward, fact-checked against source (Step 4). This is the common case (also use it after a major tool or role refactor). - From-scratch seed mode —
CLAUDE.mdis absent. There is no reservoir; every section's content must be built from the grounding sources named in Steps 3–4 (source files, onboarding seeds, observed codebase patterns) plus the per-section content checklists in Step 5. Steps that read or diff the old file — Step 2, and the "carry forward" half of Step 4 — are skipped. Where a micro-convention has no file or seed source, infer it from the dominant pattern in the codebase and mark itinferredin the Step 8 diff so the user can confirm or correct it.
State the selected mode at the top of the run.
Step 1: Measure
Compute current sizes, measuring only files that exist (in from-scratch seed
mode neither file is present, and a new repo may lack AGENTS.md even in
refactor mode — a bare wc CLAUDE.md AGENTS.md would exit non-zero):
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.
- 1mo ago First seen · 1,098 lines · 203 tokens per session scan A b11c2dfd508e
update-agent-context is a skill published in the GitHub repository mostlyharmless-ai/watercooler (14 stars, last pushed 3mo ago), licensed Apache-2.0. It adds 203 tokens to every session and 13,698 once invoked, about $0.0008 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 2026-08-30.
Other skills, from other repositories
moai-foundation-context
Manages context window optimization, session state persistence, and token budget allocation for multi-agent workflows. Use for token budget management, context limits, or session handoff across agents.
state-management
STATE.md reading, writing, and field-level updates. Provides cross-session state persistence via .planning/STATE.md with structured fields for current task, completed phases, blockers, decisions, and quick tasks.
mem0-integration
Mem0 memory layer integration for AI agents. Implement persistent, semantic memory for long-term context retention and personalization.
context-preservation
State capture and restore across context window compactions. Monitors usage thresholds and serializes quality, task, and spec state for seamless continuation.
persistent-memory
Observation capture and retrieval across sessions. Stores decisions, discoveries, and bugfix patterns. Searchable via tags and relevance scoring.
context-management
Project context loading, isolation, and persistent state management across CCPM sessions.