Skip to content

cli-reference.md

Latest commit

 

History

History
392 lines (261 loc) · 8.63 KB

cli-reference.md

File metadata and controls

392 lines (261 loc) · 8.63 KB
Raw

CLI Reference

All commands are available as wednesday-skills <command> or the short alias ws-skills <command>.

Setup Commands

install

wednesday-skills install [dir] [--skip-config] [--all]

Installs skills and configuration into a project.

  • dir — target directory (default: current directory)
  • --skip-config — skip the interactive API key prompt
  • --all — install all optional integrations without prompting

What it creates:

File Purpose
.wednesday/skills/ All skill markdown files
CLAUDE.md Claude Code skill instructions
GEMINI.md Gemini CLI skill instructions
.cursorrules Cursor editor rules
.github/copilot-instructions.md GitHub Copilot instructions
.wednesday/hooks/ Git hook scripts

config

wednesday-skills config

Interactive wizard for configuring API keys and model preferences. Writes to .env.

sync

wednesday-skills sync [--tool <name>]

Re-injects updated skill content into all tool config files. Run after upgrading the package or adding new skills.

  • --tool claude — sync only Claude Code config
  • --tool gemini — sync only Gemini CLI config

Intelligence Commands

map

wednesday-skills map [dir] [--full] [--incremental] [--ignore <pattern>] [--report-only]

Parses your source files and builds the dependency graph.

  • --full — full rebuild (ignores cache)
  • --incremental — only re-parse files changed since last run (default when cache exists)
  • --ignore <pattern> — glob pattern to exclude (can be repeated)
  • --report-only — print stats without writing files

Output files:

File Description
.wednesday/graph.db SQLite database (primary store)
.wednesday/codebase/dep-graph.json Human-readable graph JSON

After running, a token cost report is printed showing actual spend vs Claude Sonnet baseline.

analyze

wednesday-skills analyze [dir] [--incremental] [--full] [--git-history]

Runs the full analysis pipeline on top of the graph.

  • --incremental — only update changed files
  • --full — rebuild everything
  • --git-history — include git commit history signals in risk scoring

Output files:

File Description
.wednesday/codebase/analysis/comments.json Tagged comments per directory
.wednesday/codebase/analysis/legacy-report.json God files, circular deps, tech debt
.wednesday/codebase/analysis/insights.json Codebase-wide metrics

summarize

wednesday-skills summarize [dir]

Generates LLM summaries for every file and writes MASTER.md. Requires an API key.

Output files:

File Description
.wednesday/codebase/summaries.json One-sentence summary per file
.wednesday/codebase/MASTER.md Full per-file docs grouped by directory
.wednesday/codebase/GUIDE.md Architecture guide with primary flows
.wednesday/codebase/SUMMARY.md Top-level codebase statistics

Summaries are cached by file hash + exports hash. Only changed files are re-summarized on subsequent runs.

fill-gaps

wednesday-skills fill-gaps [dir] [--file <pattern>] [--min-risk <num>] [--silent]

Resolves dynamic patterns the static parser missed (dynamic requires, event emitters, global injections). Calls Haiku with a 50-token cap per gap. Edges are only added when confidence ≥ 0.80.

  • --file <pattern> — only fill gaps in matching files
  • --min-risk <num> — only process files with riskScore above threshold (default: 50)
  • --silent — suppress per-gap output

onboard

wednesday-skills onboard [dir]

Interactive developer onboarding. Asks guided questions and generates a reading guide scoped to the layer you need to touch.

chat

wednesday-skills chat "<question>" [dir]

Answers codebase questions using the pre-computed graph. Most questions return answers with zero LLM calls.

wednesday-skills chat "What calls userService?"
wednesday-skills chat "What breaks if I change auth.ts?"
wednesday-skills chat "Who wrote the payments module?"
wednesday-skills chat "List all files with risk above 80"

Analysis Commands

blast

wednesday-skills blast <file> [dir]

Shows the blast radius for a file — all files that would be affected if it changed.

wednesday-skills blast src/core/auth.ts

Output:

Blast radius: src/core/auth.ts
  Direct dependents:  4
  Transitive:        18
  Cross-language:     2
  Files: src/api/routes.ts, src/middleware/guard.ts, ...

score

wednesday-skills score <file> [dir]

Shows the risk score (0–100) for a file.

wednesday-skills score src/payments/processor.ts

Output:

Risk score: src/payments/processor.ts
  Score: 84 / 100  [Critical]
  Dependents:     32 (×1.2 weight)
  Public contract: yes (+25)
  Test coverage:   12% (+12)
  Bug-fix commits: 4 (+12)
  Tech debt:       high (+15)

  Action: Do not edit without full blast-radius review and test coverage above 60%.

dead

wednesday-skills dead [dir]

Lists dead code: files with no importers (and not an entry point) and exports with no usages.

wednesday-skills dead

Output:

Dead files (7):
  src/utils/legacy-formatter.js   (0 importers)
  src/helpers/old-validator.ts    (0 importers)
  ...

Orphaned exports (12):
  src/api/helpers.ts  →  formatDate (unused)
  ...

legacy

wednesday-skills legacy [dir]

Generates a legacy health report: god files, circular dependencies, tech debt zones, unannotated dynamic patterns.

drift

wednesday-skills drift [dir] [--rule <name>] [--since <commit>] [--fix]

Validates the codebase against architecture boundaries defined in PLAN.md.

  • --rule <name> — check only a specific rule
  • --since <commit> — only flag violations introduced after this commit (for PR checks)
  • --fix — generate suggested fixes for each violation
wednesday-skills drift --since HEAD~5    # check the last 5 commits

api-surface

wednesday-skills api-surface <file> [dir]

Distinguishes exports that are part of the public contract (imported by other modules) from internal-only exports.

trace

wednesday-skills trace <file> [function] [dir]

Traces call chains from a file or specific function.

wednesday-skills trace src/api/routes.ts handleCreateOrder

stats

wednesday-skills stats [dir]

Shows skill utilization metrics and historical LLM token costs from .wednesday/token-log.json.

Code Quality Commands

gen-tests

wednesday-skills gen-tests [dir] [--min-risk <num>] [--max-files <num>]

Generates test files for uncovered high-risk modules. Selects targets where riskScore > min-risk and testCoverage < 30%, sorted by riskScore × (100 - coverage). Requires Sonnet-level API access.

  • --min-risk <num> — minimum risk score to target (default: 50)
  • --max-files <num> — cap on files to generate tests for (default: 5)

Registry Commands

list

wednesday-skills list [dir]

Lists all installed skills with descriptions.

search

wednesday-skills search <term>

Searches the community skill registry.

add

wednesday-skills add <skill-name>

Installs a skill from the registry into .wednesday/skills/.

update

wednesday-skills update [skill-name]

Updates all installed skills (or a specific one) to the latest version.

Dashboard

wednesday-skills dashboard [--pr <num>]

Launches an interactive terminal UI (built with Ink/React) for:

  • Open PR tracking with review status

  • Unassigned semantic fix queues

  • Installed skills status

  • Historical LLM token cost breakdown

  • --pr <num> — jump directly to a specific PR

Requires GITHUB_TOKEN in .env for PR data.

Environment Variables

Variable Purpose
OPENROUTER_API_KEY OpenRouter API key (preferred)
ANTHROPIC_API_KEY Anthropic API key (fallback)
OPENROUTER_MODEL_HAIKU Model for fast/cheap calls (default: google/gemini-2.5-flash-lite)
OPENROUTER_MODEL_SONNET Model for heavy tasks (default: google/gemini-2.5-flash)
ANTHROPIC_MODEL_HAIKU Anthropic model for fast calls (default: claude-haiku-4-5-20251001)
ANTHROPIC_MODEL_SONNET Anthropic model for heavy tasks (default: claude-sonnet-4-6)
GITHUB_TOKEN GitHub API token for dashboard PR data