Triage feedback

Skill designgrappler/agent-os/claude/skills/triage-feedback

Agent OS is a management layer for complex, multi-agent AI projects. Instead of letting agents get lost in endless conversational loops, it enforces a disciplined workflow: upfront planning, shared context, scoped work tracks, and strict quality gates — so every agent stays focused, coordinated, and aligned from start to finish.

Install

npx -y skills add designgrappler/agent-os --skill triage-feedback

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Read-only triage of open Agent OS feedback issues on designgrappler/agent-os. Groups issues by subject cluster and outputs a sprint-planning summary. Invoked by /start-sprint or explicit user command.

SKILL.md

2.7 KB, as published. Nobody here has run it

Triage Feedback

Read-only triage of open feedback-labeled issues on designgrappler/agent-os. Groups by subject cluster and outputs a markdown summary for sprint planning.

Trigger

Invoked in two ways only:

Explicit user command: /triage-feedback, "triage feedback", "review feedback issues", "what feedback is open"
Programmatic invocation by /start-sprint as part of its information-gathering phase.

Auto-loading on free-text planning phrases is explicitly disabled (disable-model-invocation: true in frontmatter). This skill never auto-loads from general chat.

Phase 1 — Fetch

Run:

gh issue list --repo designgrappler/agent-os --label feedback --state open --json number,title,body,createdAt,author --limit 100

Capture the result as FEEDBACK_ISSUES.

If the command errors (auth failure, network issue), surface the gh stderr verbatim and stop. If /start-sprint is the caller, return the error string for its non-blocking handler.

Phase 2 — Group by theme

For each issue, extract the SUBJECT field from the issue body. The /submit-agent-os-feedback template embeds it as:

**Subject:** <value>

Parse the **Subject:** line. If absent or not parseable, classify as uncategorized.

Group issues by subject value. For each cluster:

Count of issues
List of issue numbers
One-line synthesis of the common pattern (inferred from titles and bodies)

Phase 3 — Triage summary output

Display a markdown report:

## Open Feedback (designgrappler/agent-os)

Total: <N> issues

### By subject

- **<subject>** (<count> issues) — #<num>, #<num> — common: <one-line pattern>
- **uncategorized** (<count> issues) — #<num>, #<num>

### Suggested track candidates

- <one-line proposal per cluster, framed as a Conductor-facing suggestion>

Conductor: which clusters become Sprint <N+1> tracks?

If FEEDBACK_ISSUES is empty, output:

## Open Feedback (designgrappler/agent-os)

No open feedback issues.

Phase 4 — Wait for Conductor decision

This skill is read-only. It produces a triage report only. The Conductor decides which clusters become tracks for the next sprint and instructs the Architect from there.

Hard Constraints

Read-only. Never invoke gh issue close, gh issue edit, or gh issue comment.
Never write to docs/context/. Output is chat-displayed text only — not a written artifact.
Never modify any file in response to this skill's invocation.