Tasuku
Workflows

Feedback and learnings

Classify review feedback and approve reusable agent guidance.

Tasuku separates observed feedback from active guidance. An agent can propose a learning, but only an administrator or super user can approve and enable it.

Finding feedback

Reviewers can classify an unhelpful finding as:

ReasonUse when
FALSE_POSITIVEThe reported defect is not present.
NOT_ACTIONABLEThe observation may be true but is not useful to change.
DUPLICATEAnother finding already covers the problem.
UNCLEARThe wording or evidence is insufficient to act on.
OTHERThe reason does not fit the standard categories.

Feedback is retained with review and repository provenance. Analytics can use it to show acceptance and noise patterns without deleting the original finding.

Learning proposals

Only explicitly mentioned pull request conversations can propose a learning. Tasuku stores the source actor, repository, interaction, content, and state. It also records descriptive metadata from the review context—including language, file path, pull request number, and source comment—so the learning can be searched and presented in other views without losing its provenance.

The repository remains the applicability boundary for repository-scoped learnings. Language is descriptive metadata; it does not make a learning apply to other repositories automatically.

Proposals do not automatically alter later prompts. This prevents an agent or one comment from silently changing organization-wide behavior.

Approval and state

An ADMIN or SUPER_USER can:

  1. inspect the proposal and its provenance;
  2. approve it;
  3. enable or disable the approved learning.

Disabled learnings remain available for audit but are not applied to new work. Re-enabling is explicit.

Repository context clusters and secondary-repository relationships follow the same manual-control principle. Suggested context is not trusted merely because an agent inferred it. Approve the relationship and keep secondary repositories read-only.

On this page