Capture per-worktree execution observations and export them to JSONL logs.
npx skills add https://github.com/espensev/ai-skills --skill observer-test
Please help me install this Agent Skill. Skill: observer-test Source: https://github.com/espensev/ai-skills/tree/main/claude-skills/skills/observer-test Command: npx skills add https://github.com/espensev/ai-skills --skill observer-test
Captures per-worktree execution observations during agent runs and exports structured data for promotion to the main project logs.
Run the observe-test start --format json command in the current worktree to begin collecting observations.
High-intent search queries and answers about installing and using this skill.
You can capture per-worktree execution observations by running the observe-test start --format json command in your current worktree, which records structured logs for tests, builds, and blockers locally in observations.jsonl.
A worktree execution observer tracks agent progress and diagnoses issues by recording test results, build errors, and blockers within isolated worktrees, exporting the structured data to a JSONL log for later promotion and analysis.
No, you do not need an external database to store agent worktree logs. The observer stores observations locally in an observations.jsonl file, supporting optional hook integration for workflows without slowing execution.
Yes, you can integrate worktree observation logging with existing agent workflows using optional hook integration. It captures machine-readable logs for tests, builds, and blockers without slowing execution in isolated worktrees.
The best way to export structured worktree test results is to use a lightweight observer that writes machine-readable logs to observations.jsonl. This allows later promotion of per-worktree data to the main project logs for analysis.
thkim2015Manage Git worktrees for isolated parallel SPEC development with automatic branch registration.
mindcockpit-aiDetect errors, warnings, and anomalies in project logs and build artifacts.
luicabref97Validate prompts with RED-GREEN-REFACTOR cycles using isolated subagents.
hilbertpRecord team observations into role-specific LEARNING.md or project DEBRIEF.md files.
