investigate-first
An agent skill by JuliusBrussee, from JuliusBrussee/caveman. Tags: analysis, debugging, troubleshooting, validation.
What it does
Diagnose ambiguous failures before editing. Use for unknown causes, intermittent behavior, performance regressions, or investigations needing evidence-ranked hypotheses.
Install
With the skills CLI, which installs into Claude Code, Codex, Cursor and other agents:
npx skills add JuliusBrussee/caveman --skill investigate-firstOr copy the skill folder into Claude Code's skills directory by hand (~/.claude/skills for every project, or .claude/skills inside one):
git clone --depth 1 https://github.com/JuliusBrussee/caveman
cp -r caveman/skills/investigate-first ~/.claude/skills/investigate-firstSafety box score
Grade A, minimal risk: 0 of 100 risk points, rated 2026-10-16. The grade rates what the skill and its scripts can reach on the machine of whoever installs it, across eight categories from shell execution to secrets access. How the score works.
This skill is documentation-only: it gives the agent a workflow for diagnosing failures before editing code, and the only files present are a markdown guide and a small YAML metadata file. It does not ship commands, scripts, network calls, or filesystem/privilege-changing behavior, so the main risk is just the agent spending more time investigating before making changes.
| Category | Level | Why |
|---|---|---|
| Code & shell execution | 0 (none) | No scripts, shell commands, or code evaluation are shipped; only `SKILL.md` and `agents/openai.yaml` are present. |
| Network access | 0 (none) | There are no download, upload, or remote-host calls in the skill files. |
| Filesystem reach | 0 (none) | The files describe a process but do not read, write, or delete any paths themselves. |
| Secrets & credentials | 0 (none) | Nothing in the skill asks for tokens, keys, passwords, or other credentials. |
| Privilege & persistence | 0 (none) | No sudo, service installation, shell startup edits, or other privilege-changing actions appear in the files. |
| Instruction hijack surface | 0 (none) | The content is a benign workflow prompt and does not tell the agent to ignore rules, hide actions, or follow external instructions. |
| Irreversible actions | 0 (none) | It explicitly says, "Do not edit until task authorizes implementation," so it reduces autonomous action rather than increasing it. |
| Transparency | 0 (none) | There are no binaries, bundled blobs, downloaded installers, or generated code; all content is plain text. |
Before installing
- No special pre-install checks are needed; this skill is documentation-only.
- If you want this behavior, confirm that you want the agent to investigate before editing on ambiguous failures.
Rated 2026-10-16 by an automated review of 2 files from the skill's repository.
Source
- Repository
- JuliusBrussee/caveman (all skills from this repository)
- Path
- skills/investigate-first/SKILL.md
- Branch
- main
- Updated
- 2026-10-05
Related skills
- principle-fix-root-causes — Apply when debugging.
- blast-radius — Find what a change could break somewhere else before it ships, beyond the diff.
- debug-voice — Use when the user runs /debug-voice, says voice mode has flaws, asks to see or capture what happened in a Grok realtime voice session.
- expo-data-fetching — Framework (OSS). Use when implementing or debugging ANY network request, API call, or data fetching.
- fix-ci — Find failing PR checks, inspect logs or external check links, and apply focused fixes
- fp-check — Systematically verifies suspected security bugs to eliminate false positives.