Skip to content

Command Reference

Matt Konda edited this page Sep 14, 2026 · 5 revisions

Command Reference

Every conscience command with its full flags and usage.

conscience setup

Check which integrations are configured and show directions for anything missing.

Usage: conscience setup

No flags. Checks: GitHub auth, conscience.yaml, Claude Code logs, dashboard endpoint, GitHub Actions workflow.

See: Getting Started


conscience examine

Full ethical analysis: signals, scorecard, and reflection questions.

Usage: conscience examine [OPTIONS]

Options:
      --repo <REPO>        GitHub repository (owner/repo)
      --days <DAYS>        Number of days to look back for GitHub data [default: 30]
      --project <PROJECT>  Project directory to filter AI logs
      --json               Output as JSON instead of formatted text

At least one of --repo or --project is required. Both can be used together for the richest analysis. Loads conscience.yaml from --project (or current directory) if present.

See: Getting Started


conscience evaluate

Ethical analysis scoped to a single pull request.

Usage: conscience evaluate [OPTIONS] --pr <PR>

Options:
      --pr <PR>            GitHub PR URL or short form (owner/repo#N)
      --project <PROJECT>  Project directory to enrich with AI session context
      --json               Output as JSON

Fetches the PR's metadata (additions, deletions, changed files, review comments) and commits, then runs the full analysis pipeline. The --project flag optionally enriches with AI session data.

See: Evaluating a Pull Request


conscience reflect

Generate reflection questions for a team retrospective.

Usage: conscience reflect [OPTIONS]

Options:
      --repo <REPO>        GitHub repository (owner/repo) to enrich questions
      --days <DAYS>        Number of days to look back [default: 30]
      --project <PROJECT>  Project directory to filter AI logs
  -i, --interactive        Answer each question at a prompt
      --json               Output as JSON instead of formatted text

Works with no data at all — questions stand on their own. Data enriches them.

Interactive mode (-i): multi-line answers (blank line finishes); Enter skips; session summary at end. With -i --json, the answered session is emitted as JSON.

See: Running a Team Retrospective


conscience attention

Analyze attention patterns across projects.

Usage: conscience attention [OPTIONS]

Options:
      --days <DAYS>        Number of days to look back [default: 7]
      --project <PROJECT>  Filter to a specific project directory
      --json               Output as JSON
      --html <HTML>        Write an HTML timeline visualization to this path

Reads Claude Code session logs only (no GitHub data needed). Default window is 7 days (shorter than other commands because attention patterns are more legible weekly).

See: Understanding Your Attention Patterns


conscience authorship

Analyze who is writing code vs. who is operating AI tools.

Usage: conscience authorship [OPTIONS] --repo <REPO>

Options:
      --repo <REPO>        GitHub repository (owner/repo)
      --days <DAYS>        Number of days to look back [default: 30]
      --project <PROJECT>  Project directory to match AI sessions
      --json               Output as JSON

Cross-references GitHub commit timestamps with Claude Code session timestamps to calculate an AI authorship correlation percentage per contributor.


conscience examine-all

Cross-project ethical analysis across all Claude Code projects.

Usage: conscience examine-all [OPTIONS]

Options:
      --days <DAYS>  Number of days to look back for GitHub data [default: 30]
      --json         Output as JSON instead of formatted text

Scans all projects in ~/.claude/projects/, runs analysis on each, and identifies outliers (token consumption, AI dependency ratio, security signals).


conscience report github

Formatted summary of GitHub activity.

Usage: conscience report github [OPTIONS] --repo <REPO>

Options:
      --repo <REPO>  GitHub repository (owner/repo)
      --days <DAYS>  Number of days to look back [default: 30]

Shows commits, PRs, contributors, review patterns.


conscience report ai

Summary of AI tool usage.

Usage: conscience report ai [OPTIONS]

Options:
      --tool <TOOL>        Filter to a specific AI tool
                           (claude-code, copilot, cursor, codex, windsurf, openclaw, nanoclaw)
      --project <PROJECT>  Filter to a specific project directory

Shows sessions, tokens, tools called, files touched. Currently only claude-code is implemented.


conscience report energy

Estimated energy consumption of AI usage.

Usage: conscience report energy [OPTIONS]

Options:
      --project <PROJECT>  Filter to a specific project directory
      --days <DAYS>        Number of days to look back [default: 30]
      --json               Output as JSON

Per-model breakdown with uncertainty ranges, total kWh, laptop-hour equivalent, CO2 estimate. All figures are estimates based on third-party research.

See: Measuring Energy Cost


conscience push

Push analysis results to a dashboard server.

Usage: conscience push [OPTIONS]

Options:
      --repo <REPO>          GitHub repository (owner/repo)
      --days <DAYS>          Number of days to look back [default: 30]
      --project <PROJECT>    Project directory
      --endpoint <ENDPOINT>  Dashboard endpoint URL (overrides config)

See: Tracking Trends with a Dashboard


conscience ingest github

Raw JSON export of GitHub data (for scripting).

Usage: conscience ingest github [OPTIONS] --repo <REPO>

Options:
      --repo <REPO>  Repository in owner/repo format
      --days <DAYS>  Number of days to look back [default: 30]

Outputs the same RepoSummary JSON that examine uses internally. Pipe to jq, save to a file, or feed to other tools.


conscience ingest claude-code

Raw JSON export of Claude Code session data.

Usage: conscience ingest claude-code [OPTIONS]

Options:
      --project <PROJECT>  Filter to a specific project directory

Outputs AiUsageSummary JSON. Includes per-session details: tokens, turns, tools used, files touched.


Global patterns

All commands that accept --project also load conscience.yaml from that directory (or the current directory if --project is not specified). See Configuration.

All commands that accept --json emit their full data structure as pretty-printed JSON, suitable for piping to jq or saving to files.

All commands that accept --repo require GitHub authentication. See Getting Started for setup.

Home

The eight commands

  • setup — what's configured
  • examine — analyze a project, PR, or all projects
  • report — github · ai · energy · tokens · authorship · attention · history · automation
  • reflect — retrospective questions
  • retro — aggregate saved reflections
  • push — send a snapshot to a dashboard
  • prune — remove a launcher behind failing automation
  • du — what conscience takes up on disk; --tidy old snapshots

Command Reference

Guides

Reference

Clone this wiki locally