A Skill directory (SKILL.md, nine modules, four references) that tells an agent to pick a fast, full or loop pass and verify its work. An optional standard-library Python controller, jspace.py, writes task state (ledger, checkpoints, open questions) to a .jspace/ folder. It installs by copying j-space/ into a host's Skills directory.
Who it is for: Users of Skill-capable AI agents running long, multi-step tasks
Strengths
Selective loading: the fast pass loads nothing extra, modules load only when needed
Optional controller uses only the Python standard library and writes only under .jspace/
Ships verify_suite.py, controller unit tests and a three-platform CI workflow
Works without a Skill loader by supplying SKILL.md as a system instruction
Weaknesses
Benchmark gains are labeled estimated from limited controlled experiments, with no method given
Benchmark scores for the listed models cannot be checked from the README
Cross-model reproduction claim comes without published results
Detected license is NOASSERTION, though the README states Apache-2.0
cavemanShortens coding-agent replies and compresses tool output before the model reads it
Live demo ↗
(opens in a new tab)119.3k stars+8,422 (+7.6%) in 30 daysApache-2.0