Tooluck logoTooluck
SKILL.md

caveats-and-status

Agent skill fromccplugins/awesome-claude-code-plugins

Use when about to report any nontrivial task as complete — requires closing with a Caveats & status section stating what was verified, what was assumed, and what's unresolved.

How to install caveats-and-status

install caveats-and-status
# install the awesome-claude-code-plugins pack with the skills CLI
npx skills add ccplugins/awesome-claude-code-plugins

# or add only this skill to Claude Code: copy the folder

plugins/craftsman/skills/caveats-and-status/

# into

~/.claude/skills/caveats-and-status/ (all projects)

.claude/skills/caveats-and-status/ (one project)

plugins/craftsman/skills/caveats-and-status/SKILL.md

Apache-2.0Full file
1# Caveats & Status
2
3## Overview
4
5"Done" without a status check is a guess wearing a confident voice. This skill defines the closing
6format every nontrivial completed task uses, so what wasn't verified is always visible rather than
7buried in an upbeat summary.
8
9**Core principle:** state what you checked, not just what you did.
10
11## When this applies
12
13Any response that reports a task as complete, fixed, or passing — code changes, file scaffolding,
14investigations with a conclusion. Skip it for pure Q&A with no action taken, and for the `quick`
15flow's trivial one-liners where the verification step itself already says everything needed (don't
16pad a two-line change with a boilerplate section that adds nothing).
17
18## The format
19
20End the response with:
21
22```
23**Caveats & status**
24- Verified: <what you actually ran/observed, with the result>
25- Assumed: <anything taken on faith because it couldn't be checked here>
26- Not covered: <what's explicitly out of scope or unverified>
27```
28
29Omit a line entirely if it's genuinely empty (e.g. "Assumed: none" is fine to state plainly, but
30don't stretch to fill a line that has nothing real to say).
31
32## Rules
33
34- "Verified" requires a command actually run or a behavior actually observed in this session —
35 not "should work," not a previous run, not extrapolation from a partial check.
36- If verification wasn't possible (no test environment, no way to run the app), say so explicitly
37…

Excerpt shown under the repository's Apache-2.0 license.

More skills in awesome-claude-code-plugins

Keep exploring

Similar skill repositories

thedotmack /

claude-mem
116k

Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More

CCCXGHGM+135 skills

Newsletter

Join the Community

Subscribe to our newsletter for the latest news and updates