Skip to main content
skill CX Codex

codependix-triage — Claude Code skill: what it does, how to install

codependix-triage — a Claude Code skill to diagnose codependix build failures and architecture drift. Triggers on failing codependix-bound targets, boundary rule violations, or stale diagram exports.

MIT 363 installs updated 2026-11-13

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add organizzolini/codebase --skill codependix-triage -a claude-code
Open on GitHub

What codependix-triage does

This skill categorizes failed codependix executions, separating boundary rule violations from stale diagram exports and project crashes. When --check boundaries flags a forbidden dependency or cycle, the edge exists in source code. Re-running the command never clears this finding: the dependency must be decoupled in code or the rule intentionally adjusted in configuration/codependix.config.ts.

For stale exports caught by --check reports, the solution is running --write to refresh the drift. Manual edits between the codependix:start and codependix:end anchor tags are lost on the next generation run.

When NestJS projects fail during preview-mode container exploration, the crash points to top-level exceptions or circular value imports evaluated during module registration.

What it helps with

  • debug architectural boundary violations and dependency cycles
  • refresh stale Mermaid diagram exports in Markdown docs
  • trace NestJS preview-mode container bootstrap failures
  • resolve failing codependix-bound targets in an Nx workspace

Example prompts for the agent

  • The codependix-bound target failed on the auth project, help triage the output.
  • Check why codependix --check reports says diagram exports are stale.
  • Help resolve a NestJS preview container boot error during codependix analysis.

How to install codependix-triage

1Skills CLI (Vercel)

npx skills add organizzolini/codebase --skill codependix-triage -a claude-code

2Manually: put the files in place

~/.claude/skills/codependix-triage/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/codependix-triage/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about codependix-triage

How to install codependix-triage in Claude Code?

Run npx skills add organizzolini/codebase --skill codependix-triage -a claude-code.

What does codependix-triage do?

It diagnoses codependix failures by distinguishing actual dependency boundary breaks from out-of-sync diagram blocks and module registration errors.

How do you fix a boundary violation reported by codependix?

Remove the unwanted dependency in the source files, lift shared types, or update the rule definitions in codependix.config.ts.

Source text

SKILL.md · MIT
---
name: codependix-triage
description: Act on a codependix run that failed or reported drift — a boundary violation, a --check reports run that found stale exports, a project that failed its export, a missing anchor block, a NestJS container that failed to boot, a rejected command line, or a --write that produced no files at all. Use when codependix exits non-zero, when a codependix-bound target fails, when an edge or a cycle breaks a declared rule, when a committed Mermaid block disagrees with a fresh run, when a graph export is missing for one project but not others, or before hand-editing an exported diagram or loosening a rule to make a check pass.
license: MIT
---

# Acting on a codependix failure

Codependix reports a handful of distinct outcomes, and reading which one
occurred is most of the work. Start by separating them: a **boundary
violation** is a real edge in real code, a **stale export** is drift, a
**failure** is a project that never got as far as producing one, and a run that
wrote nothing at all is usually neither.

## A boundary violation

`--check boundaries` found an edge, or a cycle, that a declared rule
condemns. Each violation names the graph level, the scope it was found in, the
rule, both endpoints, the project or projects it is charged to, and whatever
the rule says about why it exists.

**Start from the project the failure names.** In an Nx workspace each project
has its own `codependix-bound` target, so a red `nx run <project>:codependix-bound`
means a finding is charged to that project: a forbidden edge to the project
owning its source, a cycle to every project on it. Reproduce it with the same
target, or with `codependix map --check boundaries --projects <project>
--config configuration/codependix.config.ts`, adding `--format json`
to read the findings as data. The finding that fails a
gate is the one charged to its project; a line reading "in dependency X, not
failing" is a note, and X's own gate is the one that fails. Fix it where the
note says it lives, not in the project that printed it.

**This is the one finding a re-run never fixes.** Nothing is stale and nothing
needs regenerating: the edge is in the code, and one of two things has to give.

1. **Remove the edge.** Usually the right answer, and usually smaller than it
   looks — move a type into a `*.types.ts` file, invert a dependency, or lift
   a shared piece into a package both ends may reach.
2. **Change the rule**, when the rule is what is wrong. That is a deliberate
   decision to write down in `codependix.config.ts`, with the reason, not a
   quiet widening of a selector until the run goes green.

Never narrow a selector, add an exclusion, or delete a rule purely to make a
failing check pass. A gate loosened to pass is a gate that no longer protects
anything, and the next reader has no way to tell it was ever meant to.

If the violation is at the `nx` level and mentions an edge with no import
statement behind it, that is the point of the level: an
`implicitDependencies` entry is a real project-graph edge, and
`@nx/enforce-module-boundaries` cannot see it. Remove the entry, or the rule
has to change.

## A stale `--check reports`

One or more configured exports disagree with a freshly built graph. The run
names the projects.

**Re-run `--write`. That is the entire fix.** The graphs move with the
workspace they describe, so a project graph that changed makes every export
that drew it stale — which is drift to refresh, not a defect to investigate.

**Never hand-edit inside the anchor markers.** The next `--write` replaces the
block wholesale, so an edited diagram is a diff that silently disappears on the
following run. The same goes for a standalone Markdown file: codependix owns
its whole contents.

Two more things a stale check does _not_ mean:

- **It is not a conformetry conformance difference.** The `codependix:start` /
  `codependix:end` markers are codependix's own syntax and carry no dependency
  on any `conformetry-*` package. Running a conformance check, or reaching for a
  conformetry generator, fixes nothing here.
- **It is not evidence of a missing anchor.** A project that has never been
  exported has no block to compare, and reports stale like any other drift.

## A project that failed

A failure is collected per project and reported by name. **One project failing
does not stop the rest** — every remaining project in that pass still runs, and
all four graph-type passes are attempted regardless of whether an earlier one
failed. So a run naming three failures is three things to fix, not one flaky
run to retry.

### The Markdown file does not exist

The only anchor problem that fails outright. A **missing anchor block** is
auto-created on `--write`; a **missing file** is not, because a project with no
readme at all is a more serious problem than an unplaced block, and inventing
the file would mean authoring a document nobody asked for.

Create the file — a project readme with at least a heading — and re-run.

### A missing anchor block, on `--write`

Not a failure. `--write` creates it, at one of exactly two safe places:

- Appended to the end of the file as a new `## 🕸️ Codependix` section, with an
  intro line and a `### <Subheading>` for the graph type.
- Inserted as a new `### <Subheading>` at the end of an existing
  `## 🕸️ Codependix` section, when an earlier graph type already created one.

It never places a block anywhere else in a document someone else is authoring,
and never duplicates the heading. If the block landed somewhere you did not
want it, move the whole marker pair by hand once and every later `--write`
replaces the content in place.

### A NestJS project that failed to boot its container

The NestJS pass explores a project's container in **preview mode**, which
registers every module and provider without instantiating any of them. A
project building a `TypeOrmModule.forRootAsync` options factory never has a
database contacted, so what breaks this pass is always something that throws
while modules are being _registered_ — a module whose static `forRoot` argument
reads a variable that is absent, a top-level throw in a module file, or a
project whose root module does not export what the pass expects. Look there.

The failure is charged to the project whose container failed to boot, and its
message names the project that owns the module or class it failed on when that
is a different one — a project importing an entity package that throws on
import fails alongside it, and the owner is where the repair goes. A
value-import cycle between entity files, where a decorator reads a class
before its module finished evaluating, surfaces here as
`Cannot access 'X' before initialization` — a boot failure, not an `acyclic`
finding.

An application is rooted in the module and export its `nestjsModules.rootModules` entry names; a
library package with no such file is rooted in a synthetic module built from
every `*.module.ts` it defines, so a module file that cannot be imported on its
own fails a library that an application would not notice.

Two fixes that look plausible and are not: starting or seeding a database (the
pass never contacts one), and building the project first (it imports the
project's own TypeScript module sources directly).

If the CLI is being run from TypeScript sources through some other loader,
check that decorator metadata survives it: NestJS constructor injection reads
metadata that plain type-stripping erases, and the failure looks like a
container that cannot resolve its own providers. The published `codependix`
binary registers a decorator-preserving loader itself, and so does the Nx gate
(through `@codependix/nx/loader`, with no `@swc-node/register` needed in the
workspace); a hand-rolled invocation has to do the same. The loader reads the
workspace root's `tsconfig.json`, so a root whose `tsconfig.json` does not emit
decorator metadata fails the same way.

### The Nx project graph could not be built

The `nxProjects` level builds one graph for the whole workspace, so when it
cannot be built the failure is charged to every judged project: each gate
reports it, and when more than one project is judged the line reads
"all N judged projects" instead of a list. It is one problem, not N. Look at
the project graph itself, not at any one project — whatever stops it from
building stops every gate the same way.

## A green run that judged less than expected

A `--projects` or `--tags` list naming one project that exists beside one that
does not still runs, and judges only the entries that matched. The run logs
`🕸️ Ignored selection entries that matched no project` on stderr, listing the
unmatched `projects` and `tags` with a hint to check the spelling, and exits as
the matching entries decide. Read stderr after a green run whose selection came
from a typed list, and fix the spelling before trusting it. A selection that
matches nothing at all is refused instead.

## A rejected command line

`--check` takes a comma-separated set drawn from `boundaries` and `reports`,
and the whole command line is read before anything else happens:

- **`--check needs a value`** — a bare `--check`, or one whose value is only
  separators. Name the set; do not drop the flag. Read as "gate nothing" it
  would be a gate that cannot fail.
- **`--check does not accept "..."`** — a name from another tool. `limits` is
  codometer's and `depth` is callidescope's; only `reports` is shared.
- **`--write cannot be combined with --check reports`** — an export cannot be
  stale in the run that just wrote it. `--write --check boundaries` is legal.
- **Naming neither `--check` nor `--write`** is asked about at a terminal, and
  refused everywhere else — including any run an agent or a CI job makes,
  since neither has a terminal to answer on. Nothing is inferred and no
  default write happens.

Every mistake on one command line is collected before any is reported, so two
complaints are two things to fix rather than two runs.

The command is `codependix map`, not bare `codependix`: a command line with no
subcommand is rejected by the argument parser before any of this applies. So is
one with no `--config <path>` (`--config configuration/codependix.config.ts`
in this repository): the flag is required, and the error names it.

## A `--write` that produced no files

The quietest outcome, and almost never a bug: the run exits 0, reports that
every configured export is current, and writes nothing, because **nothing was
configured**. Every unset `target` defaults to `"none"`, and a project
resolving to `"none"` is left out of the results entirely rather than reported
as up to date — so `0` projects in the report means codependix was configured
to produce nothing.

Work down this list before suspecting the tool:

1. **Is `--config` naming the file you meant?** The flag is required and the
   file must exist, so a missing or misspelled path fails loudly — but the
   wrong file reads exactly like an empty one.
2. **Is `target` set?** A graph type carrying destinations but no `target`
   still resolves to `"none"`.
3. **Is the field spelled `defaults`?** `default` is stripped as an unknown
   key and has no effect.
4. **Does the configuration name an `include` at all?** It defaults to
   nothing, so a configuration naming only `defaults` selects no project and
   exports nothing. The run warns — `🕸️ Selected no project to export` — and
   still exits zero, and `--check boundaries` stays green either way, because
   it judges every project regardless of `include`. Write `include: ["**"]`
   for whole-workspace coverage.
5. **Do the projects match `include`, and escape `exclude`?** Both match
   against a project's Nx name _and_ its workspace-relative root.
6. **Is `--directory` the workspace root?** It is what every project's path
   resolves against, not a subtree to scan — Nx resolves the graph itself from
   the process working directory. Pointed at a single project's folder, the
   run still reads the whole graph but resolves every project underneath that
   folder, so exports land somewhere unexpected or fail on a readme that is
   not there.

The `codependix-configure` skill covers each of those fields.

## A configuration that was rejected

Validation refuses two destination mismatches rather than silently exporting
half of what was asked for: a `"both"` or `"json"` target with no `json`
destination, and a `"both"` or `"markdown"` target with no `markdown`
destination. A `markdown` destination naming neither an `anchor` nor a `path`
is refused for the same reason — nothing would say where the export goes.

## Whose problem the graph is

A finished export is a description, not a verdict — an export gates nothing
about a codebase's shape, and only a declared boundary rule does. A diagram
that shows something alarming (a cycle, a project depending on far more than
expected) is a real finding about the code even when no rule condemns it, and
the fix belongs in the code rather than in the configuration that exported
it. Reach for the `codependix-navigate` skill to read what the graph is
actually saying before changing anything.

© organizzolini · MIT · organizzolini/codebase

Similar skills

All skills for Code review and refactoring →
nx-monorepo giuseppe-trisciuoglio A skill for Claude Code, Codex, Cursor, and other AI agents that guides Nx monorepo work in TypeScript and JavaScript projects. Covers workspace setup, project generation, affected commands, CI/CD pipelines, and Module Federation. CX Codex +1 4,000installs devkit.typescript.code-review Giuseppe Trisciuoglio The devkit.typescript.code-review command reviews TypeScript monorepos in Nx workspaces covering NestJS backends, React web apps, and React Native mobile clients. CX Codex +1 ★ 372repo stars nestjs-best-practices Kadajett nestjs-best-practices is a Claude Code skill for writing, reviewing, and refactoring NestJS code. It covers architecture, security, and performance tasks. CX Codex +1 32,000installs dependency-vulnerability-triage patricio0312rev dependency-vulnerability-triage is a Claude Code skill that turns npm audit and Snyk results into a prioritized patch plan with severity ratings, safe upgrade paths, and breaking-change analysis. CX Codex +1 296installs nestjs-architecture-principles amirtaherkhani The nestjs-architecture-principles Claude Code skill plans and reviews NestJS application architecture. It activates for module-boundary changes, dependency cycles, or decisions about CQRS and microservices. CX Codex +1 592installs dependency-triage Cobus Greyling dependency-triage scans manifests and lockfiles for outdated or vulnerable packages. Classifies update types and prepares a prioritized patch list. CX Codex +1 ★ 11,819repo stars

From the same set13

The whole set →
codependix-configure codependix-configure guides creating and editing codependix.config.ts to export dependency graphs and enforce architecture boundaries. CX Codex +1 364installs codependix-export codependix-export — a skill for Claude Code that runs codependix dependency graph exports and boundary checks. Handles report verification, boundary violations, and artifact generation. CX Codex +1 364installs codependix-navigate codependix-navigate teaches the agent to answer dependency questions: how project A reaches C, what breaks if a file changes. Uses codependix path or graphs already committed to the repo. CX Codex +1 364installs codometer-measure codometer-measure — a skill that tells the agent how to run codometer, route its reports and read the JSON back. CX Codex +1 363installs codometer-configure The codometer-configure skill edits codometer.config.ts for Codometer, covering inputs, metric limits, exclusions, and report destinations. CX Codex +1 363installs codometer-triage codometer-triage handles a failing codometer --check run: tells a limit breach apart from a stale report or a run that could not finish, and points at what to fix. CX Codex +1 363installs callidescope-trace callidescope-trace teaches the agent to run callidescope and read what comes back: a call stack from an entry point, a callable's breadth, or a single project's gate verdict. CX Codex +1 361installs callidescope-configure callidescope-configure guides Callidescope configuration across command-line flags and callidescope.config.ts files. CX Codex +1 361installs

Project sponsors

Companies that keep this analytics open