SathiaAI

KnoTrack — Development skill for Claude Code

Development community

KnoTrack: a project manager for your projects, not an orchestrator.

How to install KnoTrack

This entry records only its repository, not the path inside it, so there is no exact command to give. Open SathiaAI/KnoTrack and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What KnoTrack does

KnoTrack: a project manager for your projects, not an orchestrator. Steps into a project, assesses status against source-of-truth docs, sequences work, and catches drift — usable from Claude, Windsurf, Codex CLI, LM Studio, Goose, Hermes, or any MCP-speaking harness.

Alternatives in Development

Agentic Stack — One brain, many harnesses 2.2k ★
Chops — 1,000+ macOS app to browse, edit, and manage skills across Claude Code, Cursor, Codex, Windsurf, and Amp 1.1k ★
Obsidian Reconcile — Find and resolve contradictions in the vault — the vault maintains its own truth 277 ★

README

KnoTrack

A project manager for your projects — not an orchestrator.

KnoTrack steps into an existing software project, reads its source-of-truth documents, reports current status, helps sequence upcoming work against declared dependencies, and detects drift (work happening that isn't reflected in the plan, or happening out of sequence). It never assigns or dispatches work — it tracks, sequences, and reports. If your project already has an orchestrator, KnoTrack supports it rather than replacing it.

It speaks [MCP](https://modelcontextprotocol.io) (the 2026-07-28, stateless revision of the spec), so it works the same way from Claude Code/Cowork, Windsurf, Codex CLI, LM Studio, Goose, Hermes, or any other MCP-speaking harness — not just one vendor's tool.

Status

Pre-release, v0.1.0. 11 of the 14 planned tools are implemented and dogfooded (KnoTrack tracks its own build using itself — see [`scripts/seed-self.ts`](scripts/seed-self.ts) and `docs/ROADMAP.md`'s T1); the remaining 3 are registered with their real, TRD-accurate input schemas so `tools/list` already reflects the full surface, but each currently returns a clear "not yet implemented" error rather than doing partial work.

Every change lands through a mandatory adversarial-review gate before it's considered reviewed: deterministic checks (build, lint, typecheck, unit, integration, secrets scan, dependency audit, SAST, IaC scan, migration lint, mutation testing) plus an independent panel of reviewer models from providers uninvolved in writing the code, with a verdict computed by script — never self-assessed. The latest full run's verdict is **PASS** (13/13 gates, 5 independent reviewers, 3 high/critical findings all confirmed and fixed with regression tests, 11 lower-severity findings triaged and tracked as backlog). Review run artifacts are kept locally (`.adversarial-review/`, gitignored) rather than committed, since they can include full diffs and raw model output.