175
total weekly downloads
19 packages
@allan_nava/agent-ledger
What each coding-agent task cost, read off Claude Code's own transcripts: per session, branch or day — cost, peak context, cache hit. No API key, nothing leaves the machine.
@allan_nava/claimcheck
A Stop hook for Claude Code that blocks a claim of completion the session's own record contradicts. Deterministic: no model, no network, no key.
@allan_nava/nava-ui
A small, dark-first UI kit in plain CSS
@allan_nava/stalecheck
Verifies that what the documentation says is still true: dead links, renamed files, line citations past the end, stale versions and facts dated long enough ago to be worth re-reading. Deterministic — no model, no network.
backlogsync
BACKLOG.md as the single source of truth: a generated ROADMAP.md, a check for CI, and a one-way sync to GitHub issues and milestones. Zero dependencies.
compress-nodejs
The Compress NodeJS library provides access to the Compress API for encoding videos, restreamers
create-nextcms-app
Generate a new NextCMS application.
disclosegate
A pre-push guard against publishing internal detail: work email addresses in commit metadata and trailers, private host, service and client names, home-directory paths and file URLs — checked before the push, because a commit on a forge stays reachable by
hm-react-utils
HiWay Media React utils functions
hookgate
Calibrated, sub-second decisions in Claude Code's hooks: a PreToolUse gate on shell commands and a Stop gate on unverified completion, answered by TypeSafe's Jev with a confidence score, escalating to the human when unsure.
kafkajs-ws
KafkaJS TS - node.js realtime framework server
@nextcms/generate-new
Generate a new NextCMS application.
@nextcms/generators
Generate a new NextCMS application.
@nextcms/nextcms
An open source Next CMS solution to create and manage your own next cms.
@nextcms/utils
An open source Next CMS solution to create and manage your own API.
qrspi
Questions → Research → Spec → Plan → Implement: a phase-gated Claude Code plugin that keeps a coding agent under 40% context by writing one artifact to disk per phase.
skilltrigger
How often a Claude Code skill's description makes the model load it, on prompts that should trigger it and near-misses that should not, measured behind gates that refuse to report a number they cannot trust.
transcriptmeter
What your agent sessions cost, from the transcripts on your disk: tokens by class, cache hit ratio, peak context, estimated cost, tool output by command — for Claude Code and Codex CLI, sizes only, nothing leaves the machine.
trimhook
Tool output trimmed at the source: a PostToolUse hook for Claude Code and Codex CLI that keeps the head and tail of a long shell result, spills the whole of it to a file the model can read, and reports what it saved.