allenai Feedback
Help a user tell the Asta team how using these plugins actually went. The skill interviews the user, synthesizes a narrative report (FEEDBACK.md), optionally gathers a few agent-generated reports that illustrate the story, and submits the bundle, via asta feedback submit, to a private store the Asta team analyzes.
This is meant to be low-friction for the user, with ability to review the information being submitted.
Place feedback into a <project-root>/.asta/feedback/<slug>/ directory, where <slug> is a short, dated kebab-case name (e.g. 2026-07-30-glaciar-evolution-analysis)
This skill is a router. Pick the workflow that matches the user's intent, open its .md file in workflows/, and follow it. Do not execute a workflow from memory.
Workflows
| Name | Purpose | Detailed instructions |
|---|---|---|
| interview | Converse with the user about their experience, then write FEEDBACK.md and stage any supporting files into a submission directory. |
workflows/interview.md |
| submit | Let the user review the submission directory, then run asta feedback submit <dir> to upload it. |
workflows/submit.md |
Routing
- Explicit request. If the user names a phase ("write up my feedback", "submit the feedback"), dispatch directly.
- General "give feedback" intent. Default to interview, then offer submit at the end. Do not chain automatically — submission is user-visible.
- Submit without a report. If the user asks to submit but no
FEEDBACK.mdexists, run interview first and chain to submit on confirmation.
Related skills for Asta team members
Asta team members may use this skill to comment on the experience of using the tools.
To self-reflect on a research project or update the skill benchmarks, use research-challenge or improve-skills from the asta-dev plugin.
Out of scope
- Uploading datasets or large working files. Supporting files are small agent-generated reports that illustrate the narrative, not inputs of independent interest — the CLI enforces size gates.
- Grading research quality.