Verified against Semitexa Ultimate 2026.11.05.1529
ai commands
Every ai command, with the arguments and options the console actually defines.
ai:agent
Agent presence: join with a name and intent, see who else is working and what is being edited, leave when done
bin/semitexa ai:agent ["action"]
| option | takes a value |
|---|---|
--name |
yes |
--intent |
yes |
--repo |
yes |
--task |
yes |
--id |
yes |
--all |
no |
--json |
no |
ai:ask
Agent-facing introspection aggregator (capabilities, mechanisms, components, project, module, route, event, path, api, logs)
bin/semitexa ai:ask ["subject"]
| option | takes a value |
|---|---|
--name |
yes |
--search |
yes |
--path |
yes |
--method |
yes |
--file |
yes |
--lines |
yes |
--grep |
yes |
--level |
yes |
--since |
yes |
--around |
yes |
--context |
yes |
--list |
no |
--kind |
yes |
--id |
yes |
--area |
yes |
--state |
yes |
--json |
no |
ai:audit
Record that right or fast was audited, and what it covered, so an empty phase reads as clean and not as unread: record | list
bin/semitexa ai:audit "<action>"
| option | takes a value |
|---|---|
--phase |
yes |
--covered |
yes |
--by |
yes |
--with-framework |
no |
--json |
no |
ai:backlog
Backlog hygiene: stats + cleanup proposals (dry-run; --apply to commit)
bin/semitexa ai:backlog "<action>"
| option | takes a value |
|---|---|
--apply |
no |
--json |
no |
ai:charter
The project charter (goal, audience, problem, signs of success, constraints, non-goals, what the owner decides) and the interview that fills it: next | answer | show | check
bin/semitexa ai:charter "<action>"
| option | takes a value |
|---|---|
--field |
yes |
--value |
yes |
--source |
yes |
--none |
no |
--keep |
no |
--replace |
no |
--json |
no |
ai:context
Pack prior art + signals for a recipe (NDJSON, agent-facing)
bin/semitexa ai:context "<recipe>"
| option | takes a value |
|---|---|
--json |
no |
--module |
yes |
--trace |
yes |
ai:decision
The project owner's decisions, with scope (standing | once), expiry and consumption: record | list | show | consume | revoke
bin/semitexa ai:decision "<action>"
| option | takes a value |
|---|---|
--id |
yes |
--words |
yes |
--means |
yes |
--applies-to |
yes |
--scope |
yes |
--expires |
yes |
--by |
yes |
--act |
yes |
--all |
no |
--json |
no |
ai:epic
Manage epics for agent-driven work decomposition (start/list/show/update)
bin/semitexa ai:epic "<action>"
| option | takes a value |
|---|---|
--id |
yes |
--title |
yes |
--goal |
yes |
--status |
yes |
--scope |
yes |
--json |
no |
--trace |
yes |
ai:evidence
Keep review evidence (screenshots, traces, exports) private with a passport and an expiry; publishing one needs the operator at a terminal
bin/semitexa ai:evidence ["action"] ["target"]
| option | takes a value |
|---|---|
--kind |
yes |
--data |
yes |
--note |
yes |
--ttl-days |
yes |
--move |
no |
--to |
yes |
--allow-real-data |
no |
--dry-run |
no |
--tmp |
no |
--json |
no |
ai:invoke
Run a payload handler from CLI, skipping HTTP/auth/middleware — dev only. Use --preview to resolve the target without executing.
bin/semitexa ai:invoke
| option | takes a value |
|---|---|
--route |
yes |
--method |
yes |
--handler |
yes |
--payload |
yes |
--json |
no |
--human |
no |
--expect-field |
yes |
--preview |
no |
ai:make-it
What to do next in this project: interview on an empty backlog, the next approved task with its steps, ideas to offer, or all done
bin/semitexa ai:make-it
| option | takes a value |
|---|---|
--with-framework |
no |
--phase |
yes |
--except |
yes |
--json |
no |
ai:observe
Observatory for agents: ps (live snapshot) | tail (journal rows, --follow streams) | show --id (one process + its trace) | timeline [--id] (a page: its UI events, stream frames, re-runs)
bin/semitexa ai:observe "<action>"
| option | takes a value |
|---|---|
--id |
yes |
--mutate |
yes |
--method |
yes |
--path |
yes |
--input |
yes |
--kind |
yes |
--name |
yes |
--lines |
yes |
--follow |
no |
--duration |
yes |
--source |
no |
--json |
no |
ai:orient
Session dashboard: git + active epic + in-progress tasks + recent traces + next step. Run this first on a cold start.
bin/semitexa ai:orient
| option | takes a value |
|---|---|
--json |
no |
--human |
no |
--traces |
yes |
ai:plan
Risk-score a recipe + change set (NDJSON, agent-facing)
bin/semitexa ai:plan ["recipe"]
| option | takes a value |
|---|---|
--json |
no |
--module |
yes |
--files |
yes |
--trace |
yes |
ai:quality
Quality ledger: check metrics against their baseline, record improvements, accept a regression with a reason
bin/semitexa ai:quality ["action"]
| option | takes a value |
|---|---|
--limit |
yes |
--metric |
yes |
--reason |
yes |
--all |
no |
--json |
no |
ai:report
Report a Semitexa framework defect (and its workaround) as a GitHub issue
bin/semitexa ai:report
| option | takes a value |
|---|---|
--title |
yes |
--summary |
yes |
--evidence |
yes |
--workaround |
yes |
--package |
yes |
--repo |
yes |
--yes |
no |
--draft |
no |
--json |
no |
ai:review-graph:capabilities
Show AI-relevant project capabilities derived from the review graph
bin/semitexa ai:review-graph:capabilities
| option | takes a value |
|---|---|
--json |
no |
--markdown |
no |
--category |
yes |
--module |
yes |
ai:review-graph:context
Build relevant context for a task
bin/semitexa ai:review-graph:context "<task>"
| option | takes a value |
|---|---|
--format |
yes |
--depth |
yes |
--module |
yes |
ai:review-graph:diff
Show how the graph changed since last scan
bin/semitexa ai:review-graph:diff
| option | takes a value |
|---|---|
--format |
yes |
--module |
yes |
--base |
yes |
--path |
yes |
--fail-on |
yes |
ai:review-graph:event-trace
Trace the full lifecycle of an event
bin/semitexa ai:review-graph:event-trace "<event>"
| option | takes a value |
|---|---|
--format |
yes |
--include-code |
no |
ai:review-graph:findings
Unused classes graded by confidence, and class dependency loops
bin/semitexa ai:review-graph:findings
| option | takes a value |
|---|---|
--kind |
yes |
--min-confidence |
yes |
--module |
yes |
--format |
yes |
--no-refresh |
no |
ai:review-graph:flow-trace
Trace an execution flow end-to-end
bin/semitexa ai:review-graph:flow-trace "<flow>"
| option | takes a value |
|---|---|
--format |
yes |
--include-code |
no |
ai:review-graph:generate
Build or update the review graph from the codebase
bin/semitexa ai:review-graph:generate
| option | takes a value |
|---|---|
--full |
no |
--json |
no |
ai:review-graph:impact
Analyze the impact of changes on the codebase
bin/semitexa ai:review-graph:impact "<target>"
| option | takes a value |
|---|---|
--depth |
yes |
--json |
no |
--no-refresh |
no |
--ndjson |
no |
--context |
no |
--prompt |
yes |
--module |
yes |
ai:review-graph:intelligence
Query the project graph intelligence layer
bin/semitexa ai:review-graph:intelligence ["query"]
| option | takes a value |
|---|---|
--hotspots |
no |
--doc-gaps |
no |
--flows |
yes |
--event-lifecycle |
yes |
--intent |
yes |
--format |
yes |
ai:review-graph:module
Overview of a module with full context
bin/semitexa ai:review-graph:module "<module>"
| option | takes a value |
|---|---|
--format |
yes |
--include-events |
no |
--include-flows |
no |
ai:review-graph:query
Run ad-hoc queries against the review graph
bin/semitexa ai:review-graph:query
| option | takes a value |
|---|---|
--type |
yes |
--module |
yes |
--usages |
yes |
--dependencies |
yes |
--cross-module |
no |
--from |
yes |
--to |
yes |
--search |
yes |
--limit |
yes |
--compact |
no |
--json |
no |
--ndjson |
no |
--no-refresh |
no |
--depth |
yes |
ai:review-graph:show
Display or export the review graph
bin/semitexa ai:review-graph:show ["focus"]
| option | takes a value |
|---|---|
--format |
yes |
--output |
yes |
--allow-anywhere |
no |
--module |
yes |
--type |
yes |
--depth |
yes |
ai:review-graph:stats
Show review graph statistics and health
bin/semitexa ai:review-graph:stats
| option | takes a value |
|---|---|
--json |
no |
ai:review-graph:watch
Watch for file changes and incrementally update the graph
bin/semitexa ai:review-graph:watch
| option | takes a value |
|---|---|
--interval |
yes |
--full-on-start |
no |
ai:skills
List AI-executable skills with full metadata. Use for debugging or AI tooling.
bin/semitexa ai:skills
| option | takes a value |
|---|---|
--json |
no |
ai:state
The project state, work · right · fast: done over approved tasks per phase, for ./src, with the framework on request
bin/semitexa ai:state
| option | takes a value |
|---|---|
--with-framework |
no |
--json |
no |
ai:task
Classify a task description into a recipe + suggested make:* invocation
bin/semitexa ai:task "<description>"
| option | takes a value |
|---|---|
--module |
yes |
--json |
no |
--ndjson |
no |
--trace |
yes |
ai:trace
Durable per-task trace: start/append/show/list events across sessions
bin/semitexa ai:trace "<action>"
| option | takes a value |
|---|---|
--id |
yes |
--topic |
yes |
--recipe |
yes |
--kind |
yes |
--summary |
yes |
--payload |
yes |
--limit |
yes |
--json |
no |
ai:verify
Run the precise lint+test+module-structure subset for a diff/file list (NDJSON, agent-facing)
bin/semitexa ai:verify
| option | takes a value |
|---|---|
--files |
yes |
--git-ref |
yes |
--diff-stdin |
no |
--dirty |
no |
--all |
no |
--scope |
yes |
--trace |
yes |
--json |
no |
--impact |
no |
--list-e2e |
no |
--list-pages |
no |
ai:verify:receipt
Check an ai:verify receipt: intact, tree unchanged since the run, verdict pass
bin/semitexa ai:verify:receipt ["id"]
| option | takes a value |
|---|---|
--unread |
no |
--hours |
yes |
--commit |
yes |
--json |
no |
ai:verify:rules
Show how often each ai:verify rule had a chance and fired; dormant rules first
bin/semitexa ai:verify:rules
| option | takes a value |
|---|---|
--dormant |
yes |
--json |
no |
ai:work
Manage work tasks: start/list/show/update/resume/note (agent-facing, NDJSON)
bin/semitexa ai:work "<action>"
| option | takes a value |
|---|---|
--id |
yes |
--epic |
yes |
--title |
yes |
--recipe |
yes |
--risk |
yes |
--phase |
yes |
--status |
yes |
--scope |
yes |
--take-over |
no |
--context-ref |
yes |
--next-step |
yes |
--note |
yes |
--tail |
yes |
--trace |
yes |
--json |
no |
ai:work:approve
Approve a task or a whole epic for implementation, or return it to ideas; approved tasks and bugs count towards progress
bin/semitexa ai:work:approve
| option | takes a value |
|---|---|
--id |
yes |
--epic |
yes |
--words |
yes |
--by |
yes |
--revoke |
no |
--json |
no |