INTENTION, MADE SIMPLE
Get what you asked for
Define exactly how your Agent should tackle a problem, what it should prioritize, and how it should communicate.
Agent Traits & Workflows
typed, versioned & in-sync.
Turn your skills into shareable modules, that are easily maintained and orchestrated by you and your team.
TERMINALimplement-task @ task-001
$ ctx traits run implement-task -- task-001
progress 2/5 · running · 1m 44s · 205 tok
Draft taskreviewer·opencodeprompt(16s · 97 tok)
Implement taskworker·opencodeprompt(1m 44s · 205 tok)
Refinement taskmasterloop
Review workreviewer·opencodeprompt
Apply fixesworker·opencodeprompt
Write the commitscribe·opencodeprompt
Commit the phasemastercommand
THE FOUNDATION
Work with your Agents,
without a headache.
EXPLICIT TYPING
Types enforced at each step
All of the inputs, outputs and communication is explicitly typed.
No more inventing what Agent believes it should produce.
LOCKFILE SAFETY
No unexpected changes
Distribution and dependency management of your traits is reviewed locally. No prompt / instruction changes without your permission.
WORKFLOW ORCHESTRATION
Deterministic, where possible
Working with Agents efficiently, requires that they follow your process. Control your outer loop with explicit workflow definitions as code.
CHECK IT OUT
Define agents with code,
write it, build it & run it.
TERMINALimport @ SKILL.md
$ ctx traits import ./SKILL.md→ parsing SKILL.md→ synthesizing typed trait$STAY IN THE LOOP
Follow our journey.
Journey has just began.
Please follow us on x.com or subscribe below for updates.
Get product updates.
No marketing, unsubscribe anytime.