CLI reference

Every coding.sh command and option, generated from the CLI's own --help output at build time.

Generated at build time from coding.sh --help (version 0.1.5). Run coding.sh <command> --help for the same text in your terminal, or add --json for machine-readable metadata. Each command has its own page with its full help and its subcommands.

Global options

Every command accepts these.

global options

      --json         Print one JSON document on stdout
      --no-color     Disable color (also NO_COLOR=1)
      --debug        Show stack traces on errors
  -h, --help         Show help
      --store <URL>  Read and write this index store instead of the configured one (also CODING_SH_STORE_URL); skips auto-ingest
      --blobs <URL>  Blob store to use with --store (also CODING_SH_BLOB_URL; a postgres:// index keeps blobs in the same DB)
  -v, --version      Print the version

With --json, a command prints one JSON document on stdout and errors as one JSON line on stderr. Usage errors exit with code 2.

Commands

Capture

Command	What it does
`init`	Set up capture hooks and the MCP server for every detected agent
`doctor`	Check agents, hooks, spool, store and the OTLP port
`status`	Show capture totals and the latest session
`statusline`	Print a Claude Code status line (model, cost, worst flag), or install it (install, uninstall)
`ingest`	Move captured hook events from the spool into the index
`run`	Run any agent command and capture the session
`collector`	Start a local OTLP/HTTP receiver for agents that export traces (import)
`proxy`	Run a local Anthropic/OpenAI API proxy that logs every model call (status, stop, env)
`serve`	Production server: OTLP + native ingest, viewer, health and metrics on one port (keygen, healthcheck)
`uninstall`	Remove coding.sh hooks and MCP entries from every agent

Explore

Command	What it does
`sessions`	List captured sessions
`show`	Show a session timeline with subagents and latency
`tree`	Show a whole multi-agent run as one tree, with rolled-up totals
`replay`	Step through a session and see what it did to each file it touched
`attribution`	Which agent session wrote the lines each commit added; export them as Agent Trace (export)
`blame`	git blame with the agent session that wrote each line
`diff-calls`	Show what changed between two model calls, and why the second missed the prompt cache
`search`	Full-text search over prompts, replies and tool calls
`usage`	Tokens, cost, turns and cache hits by agent, model or day
`cache`	Score prompt-cache health and name what breaks the cached prefix (doctor)
`stats`	Sessions, failures, cost, tokens and latency per day or week, split by agent, model, repo or flag
`sql`	Run one read-only SQL query against the SQLite index
`export`	Stream sessions, events or labels as JSONL or CSV, or write raw traces to a directory
`ui`	Open the local session viewer
`mcp`	Serve captured sessions to agents over MCP (stdio)

Evaluate

Command	What it does
`eval`	Build eval suites from sessions, skills or cases, and run them (init, run, candidates, new, import, validate, crosscheck, audit, trigger, list, export)
`dataset`	Manage versioned datasets of cases (add, import, list)
`exp`	Run a task over a dataset, compare experiments and sweep configs (ab, replay, run, session, compare, regrade, aa, sweep, metrics, plateau, show, list, status)
`score`	Add and list typed scores on sessions, events, cases and runs (add, list, config, configs)
`judge`	Grade one output or the slop in captured diffs, or check an LLM judge or a test verifier against human labels (grade, slop, calibrate, sample)
`toolevals`	Measure how well agents use a CLI, MCP server or API (init, surface, validate, run, compare, compress, show, traces)
`workflows`	Capture, measure and A/B-test multi-agent workflow runs (list, import, show, stats, score, run, compare, replay)
`bench`	Dynamic benchmarks: fresh cases every run, same set for every contender (new, run, regrade, list, show, history, report, schedule)

Harness

Command	What it does
`improve`	What to fix first in your sessions, the line that fixes it, and whether your last AGENTS.md changes helped (fix, check)
`harness`	Snapshot and list agent harness bundles; audit which memory and instruction files agents use (snapshot, list, audit)
`compare`	A/B two harness bundles on eval tasks
`refine`	Improve AGENTS.md and skills in tested rounds, and protect their safety rules (propose, test, apply, rollback, log, run, protected, check, mine, recheck, distill, diagnose, signals, score)

Maintain

Command	What it does
`gateway`	LLM gateways: detect routing, enrich sessions with billed cost (status, enrich)
`gc`	Delete expired spool files, blobs and judge cache
`purge`	Remove a leaked secret from the store: events, search, blobs, spool, eval runs
`config`	Read and write ~/.coding.sh/config.toml (get, set, path)
`update`	Install the latest coding.sh release
`upgrade-store`	Migrate the store for this install, or put back the index from before a migration
`version`	Print the version