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/scrolldynasty/browsermesh/write-adrnpx skills add scrollDynasty/BrowserMesh --skill write-adrgit clone --depth 1 https://github.com/scrollDynasty/BrowserMeshWhat 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 | $0.00060 | $0.00643 |
| Opus 5 | $0.00030 | $0.00321 |
| Sonnet 5 | $0.00012 | $0.00129 |
| Haiku 4.5 | $0.00006 | $0.00064 |
Grade A, and why
write-adr 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 yesterday.
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 — 61 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Writing an ADR
ADR 0012 requires a controlled design before code for anything that changes a contract or an invariant. The ADR is that design, not a retrospective write-up.
Format
Follow ADR 0019 and 0020. They are the current form; the eighteen earlier files use two older layouts and are not the template. Do not retrofit them.
# ADR NNNN — Title in sentence case
Status: accepted
Date: YYYY-MM-DD
## Context
## Decision
## Consequences
- Filename:
docs/decisions/NNNN-kebab-case-title.md,NNNNzero-padded, next unused number. - Em dash after the number, not a colon.
Status: acceptedlowercase — alsoproposed,superseded by ADR NNNN.- Sentence case in the title. No trailing period.
What each section carries
Context. The measured problem, not the wish. State numbers where numbers exist: byte sizes,
test counts, timings, the version something was observed against. ADR 0020 opens with
"tools/list was 134,839 bytes, 7.3× the official Playwright MCP server" — that is the standard.
Name the constraint that rules options out; for BrowserMesh that is often a SPEC clause.
Decision. What is now true, in the present tense. Include what was rejected and why, especially
the option a reader would otherwise propose. If a check lives in one layer rather than another, say
which and why — ADR 0020 explains that an input-schema rejection names the offending field while a
BrowserMeshError would surface as an uninformative INVALID_ARGUMENT.
Consequences. What this costs, what it forecloses, what now has to stay true. A consequence section that only lists benefits is unfinished.
Rules
- One decision per ADR. A second decision is a second file.
- An accepted ADR is not edited to reflect a later change of mind. Write a new ADR and set the old
one to
superseded by ADR NNNN. - The ADR outranks
README.mdand the docs site but is outranked bydocs/SPEC.mdanddocs/architecture.md— see the source-of-truth order inCLAUDE.md. If the ADR contradicts the SPEC, one of them is wrong; resolve it before merging. - Update
docs/architecture.mdand any affected public tool description in the same change. Docs and implementation must not diverge silently. docs/decisions/is not in.prettierignore, so runnpm run formatbefore committing.
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.
- yesterday First seen · 61 lines · 60 tokens per session scan A 3be2c7308d9c
write-adr is a skill published in the GitHub repository scrollDynasty/BrowserMesh (4 stars, last pushed yesterday), licensed Apache-2.0. It adds 60 tokens to every session and 643 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.
Other skills, from other repositories
browser_harness
../../SKILL.md.
browser-harness
Always use browser-harness for any web interaction: automation, scraping, testing, or site/app work.
deep-reverse
Use only for explicit web or API protocol reverse engineering: endpoint discovery, request replay, signature/token/encryption analysis, JavaScript runtime tracing, browser-independent reproduction, or crawler implementation that must not depend on browser automation. Do not use for one-off visible-page extraction or…
web-crawl
Get data from websites with the least necessary complexity. Use for quick lookups, visible-page extraction, search results, tables, pagination, detail-page traversal, downloadable files (Excel/CSV/PDF/ZIP), embedded page state, internal JSON/XHR APIs, bulk collection, reusable crawler generation, and delivery in any…
eric-task-master
Install, connect, and use Eric Task Master for durable Playwright browser jobs with persistent or ephemeral Profiles, live fast-auto-human behavior control, task queues, progress health, semantic diagnostics, recovery, artifacts, and composable Task Packs.
opencli-usage
Use at the start of any OpenCLI session — this is the top-level map of what opencli can do, how to discover adapters, what flags and output formats are universal, and which specialized skill to load next. Point here when an agent asks "what can opencli do?" or "how do I find the right command?".