Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
git clone --depth 1 https://github.com/lugassawan/swe-workbenchWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/lugassawan/swe-workbench/address-feedback)<a href="https://agentmods.dev/commands/lugassawan/swe-workbench/address-feedback"><img src="https://agentmods.dev/badge/commands/lugassawan/swe-workbench/address-feedback/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/commands/lugassawan/swe-workbench/address-feedback"><img src="https://agentmods.dev/badge/commands/lugassawan/swe-workbench/address-feedback.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00046 | $0.00589 |
| Opus 5.5 | $0.00018 | $0.00236 |
| Sonnet 5.5 | $0.00009 | $0.00118 |
| Haiku 4.5 | $0.00005 | $0.00059 |
Grade A, and why
address-feedback scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 38 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Step 1 — Parse arguments
Parse $ARGUMENTS:
- Matches
[1-9][0-9]*(stripping a leading#if present) → use as$PR. - If empty, run
gh pr view --json number 2>/dev/null. If it succeeds, offer:"Detected PR #N on this branch — address feedback on it? Reply
yesto proceed." Wait for reply.yes→ use that PR number. Else → print "No PR number provided." and stop. - If non-numeric and no PR detected → print usage hint and stop.
Step 2 — Ticket context (when a ref is present)
Inspect $ARGUMENTS plus git rev-parse --abbrev-ref HEAD and git log --oneline -5 for ticket keys ([A-Z]+-\d+), atlassian.net URLs, Confluence wiki URLs, or #NNN GitHub refs. If matched, invoke swe-workbench:ticket-context and prepend its structured summary to the delegation context in Step 3.
Step 3 — Invoke workflow skill
Invoke swe-workbench:workflow-address-feedback with the resolved PR number and any ticket-context prelude from Step 2.
The skill owns: pre-flight + fetch (via swe-workbench-address-feedback-fetch), owner-identity check, per-thread A/C/D triage, Edit-tool fix application,
swe-workbench:workflow-commit-and-pr for committing, REST reply posting, GraphQL
resolveReviewThread for review threads (all three dispositions — ADDRESSED, CLARIFIED, DEFERRED —
resolve; PR-level comments have no thread to resolve), and PR-metadata drift sync (Phase 6).
Worktree handling belongs to the skill, not this command:
- Setup (Phase 2) — via
swe-workbench-address-feedback-worktree acquire, which reuses the current checkout when it is already on the PR head branch, or an existing worktree registered for that branch; otherwise creates one on the PR branch itself, so Phase 4 commits push straight to the PR. - Cleanup (Phase 7) — via
swe-workbench-address-feedback-worktree release, which always removes a worktree it created, on success,Q-quit, or error, while keeping the local PR head branch. No-ops when Phase 2 reused an existing checkout.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago First seen · 38 lines · 46 tokens per session scan A 5cece8e51057
address-feedback is a command published in the GitHub repository lugassawan/swe-workbench (3 stars, last pushed 4d ago), licensed MIT. It adds 46 tokens to every session and 589 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other commands, from other repositories
review
Fast review of staged changes. Critical findings only — no style nits.
pre-push
Run the full review skill on all changes since last push.
review-pr
Perform a comprehensive code review using subagents for key areas.
local-code-review-tests
Review uncommitted changes, build projects, and run unit tests.
local-code-review
Review uncommitted changes and build .NET/Angular projects.
git
The pre-finish status: branch, hygiene findings, message checks, workflow lint, template state.