Show whether the agent followed the kit's rules in recent sessions: pages edited and measured, Stop hook blocks, findings left open. Reads local hook events only.
Start a fresh conversation. Replace the highlighted placeholders.
When I explicitly run /observe, use observe to report kit-rule compliance for [project path] over [number of days]. First check for the kit's scripts/observe.mjs; if missing, explain that only skill folders were installed, point to the plugin or documented init command, and stop. Run the local hook-event report with the requested project/window, quote its summary lines, and stop if no session edited UI. For each BROKE A RULE session, read --session details and cite the failed rule and timeline evidence. Report only recorded hook events, not memory. Explain the three counted rules and that unfinished sessions are excluded; this does not measure design quality.
Required inputs
Project directory and requested time window
User-approved starter prompt based on a pinned source version; check current skill requirements before running. Task evidence ↗
SKILL.md
name: observedescription: 'Show whether the agent followed the kit''s rules in recent sessions: pages edited and measured, Stop hook blocks, findings left open. Reads local hook events only.'# A command the user runs, like /gate. The model never starts it.disable-model-invocation: true
Step 0 - is the kit here? Check once:
ls ${CLAUDE_SKILL_DIR}/../../../scripts/observe.mjs >/dev/null 2>&1 && echo KIT_OK || echo KIT_MISSING
On KIT_MISSING only the skill folders were installed. Say so, point the user
at the plugin install or npx ux-ui-agent-skills init, and stop.
Report what the hooks recorded - never a reconstruction from memory.
Run node ${CLAUDE_SKILL_DIR}/../../../scripts/observe.mjs --project "$CLAUDE_PROJECT_DIR"
(or without --project when the user asks about every project, and with
--days N when they name a window).
Quote the summary lines it prints. If it says no session edited UI, say that
and stop: there is nothing to judge.
For each session marked BROKE A RULE, run it again with --session <id> and
say which rule failed and the timeline line that shows it.
If the user wants the page, run it with --html <path> and give the path.
What it measures, and what it does not: three counted rules per session -
every edited page had a passing gate before the end, no fast-gate finding was
left on a file's latest version, and the agent never finished while the Stop
hook still had pages open. It says nothing about design quality, and an
unfinished session is not counted for or against.