
Skill Feedback Writer
Reviews another skill's output and writes feedback on what instruction is missing
- Repository
- topoteretes/cognee
- Stars
- 33,002
- Licence
- Apache-2.0
- Files
- 1 file
What it does
This skill makes an agent act as a reviewer of another skill's output. It judges whether that output was good enough for the task, scores it from 0.0 to 1.0, and names the specific instruction the other skill is missing. It is part of a demo feedback loop in the cognee repository, and its instructions focus on the pr-comment-evaluator skill, which is meant to compare reviewer comments against concrete runtime risks rather than judge tone alone.
Great for
- Developers building or testing agent skills who want an automated critic for another skill's output
- Researchers or maintainers exploring feedback loops for improving SKILL.md instructions
- Anyone studying how the cognee demo wires skills together
Skip it if
- General end users who want a finished deliverable: the skill only produces review JSON, not work product (reason: its output is evaluation, not content)
- Use as a sole quality gate for production decisions: the scoring rules are short and the scores are the agent's judgment, not validated metrics (reason: no tests or calibration are included)
Try asking
- “Evaluate whether the pr-comment-evaluator output caught the main runtime risk in a diff”
- “Score a skill's output and explain which instruction is missing”
- “Run the feedback loop demo to suggest improvements to pr-comment-evaluator”
Install in 3 steps
- 1
Download the skill
The ZIP contains the 1 original files, the Apache-2.0 licence and an attribution note.
skill-feedback-writer.zip - 2
Add it to your agent
Claude Code (all projects)
mkdir -p ~/.claude/skills && unzip skill-feedback-writer.zip -d ~/.claude/skills/Claude Code / Agent SDK (this project only)
mkdir -p .claude/skills && unzip skill-feedback-writer.zip -d .claude/skills/Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.
- 3
Ask for the task
No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.
Get better results
- Provide the other skill's actual output and the diff or task it was given, since the review depends on those inputs
- Read the missing_instruction field carefully; the skill is built to name a specific gap rather than give general praise
- Treat scores as a rough signal and check important judgments yourself
What's inside
- SKILL.md
Redistributed unmodified from topoteretes/cognee under the Apache-2.0 licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.