Installs a per-turn first-line canary (user name + turn counter + honest
context self-check) so attention drift, compaction, and truncation become
visible the moment they happen, with a checkpoint/re-anchor/reset trip
protocol. References file documents the research grounding: Chroma's
context-rot study, Lost in the Middle (Liu et al.), instruction-drift
findings, compaction summary loss, and Breunig's context failure taxonomy.
Updates README, CLAUDE.md, and CONTEXT.md skill rosters to six skills.
https://claude.ai/code/session_01CWPytH8cfzYfCgU6eC9pX9
Treats agent-generated plans as junior work and constructs a senior
reviewer grounded in codebase research plus web research of current
best practices. Diagnoses altitude failures (fog: vague on hard parts;
tunnel: granular with no product vision), reviews adversarially with
evidence-backed findings, and rewrites the plan into a scoped,
state-of-the-art version with product decisions surfaced as open
questions.
Includes references for the review rubric (altitude tests, vague-word
blacklist, severity calibration) and the research playbook (code and
web research tracks, source quality ranking, stop conditions).
https://claude.ai/code/session_01EXN9BXWy6933G7RNQmmUU5
Rename the `ui-craft` skill to `interface-kit`: move/rename the skill directory and reference files (skills/ui-craft → skills/interface-kit), update SKILL.md (name and heading), and update repository docs (CLAUDE.md, CONTEXT.md, README.md, docs/adr/0001-repo-shape.md) to reflect the new name and wording. Keeps priorities and content intact while standardizing naming across the repo.
Introduce the new `loop-factory` skill and its documentation. Adds skills/loop-factory/SKILL.md plus reference docs (autonomous.md, commands.md, install.md, spec-authoring.md) describing the spec-driven inbox→active→archive loop, autonomous mode, CLI commands, install steps, and spec authoring guidance. Also updates README.md, CONTEXT.md, and CLAUDE.md to list and summarize the new skill.
Claude