Get Pr ReviewsCAUTION
The AI toolkit for building reliable browser automations
Overview
The AI toolkit for building reliable browser automations
eeed53996b1dOBSERVED · 2026-10-13What it tells the agent
The instruction file, verbatim from the audited commit — this is the text the model reads, and the surface the audit's instruction layer examines. Quoted here so you can judge it without cloning anything.
---
name: get-pr-reviews
description: Get nicely formatted output of the last review comments on a PR. Used internally by other skills.
---
Get nicely formatted review comments for a pull request, grouped by file and line with proper comment threading.
## Steps
1. **Determine PR number**: If not provided, look up the associated PR for the current branch using:
```bash
gh pr view --json number --jq '.number'
```
2. **Get formatted review comments**: Use the GitHub API to fetch and format all line-specific review comments:
```bash
gh api repos/:owner/:repo/pulls/{PR_NUMBER}/comments --jq '
group_by(.path, .line) |
map({
path: .[0].path,
line: .[0].line,
comments: sort_by(.created_at)
}) |
sort_by(.path, .line) |
map(
"**\(.path)**\n Line \(.line):" +
(.comments |
map(
if .in_reply_to_id then
" | \(.user.login): \(.body)"
else
" (\(.user.login)): \(.body)"
end
) |
join("\n")) + "\n"
) |
join("\n")
'
```
## Output Format
The formatted command produces output like:
```
**apps/api/src/handlers/optionsHandlers.ts**
Line 98: (tanishqkancharla): I think this is a bug. If you search for e.g. "blue cros", none of these search conditions will match.
| tanishqkancharla: I see that you re-query all insurances below, so this doesn't get used.
| Mochael: I think we should ditch the ilikes and just use the fuzzy search applied to planName, payer name and abbreviation
| tanishqkancharla: default to fuzzy search
**apps/api/src/handlers/optionsHandlers.ts**
Line 116: (tanishqkancharla): fyi: don't need to fix, but I think you can just do `query.where(and(...conditions, or(...searchConditions)))`
```
## Key Features
- **Grouped by file and line**: Comments are organized by their location in the code
- **Threaded conversations**: Replies are indented with `|` to show the conversation flow
- **Chronological order**: Comments within each thread are sorted by creation time
- **User attribution**: Each comment shows the GitHub username of the author
## Usage Example
```bash
# Get formatted review for PR #84
gh api repos/saffron-health/monorepo/pulls/84/comments --jq '...'
```Trust audit
CAUTIONgrade B · trust 89/100 Install with care. The audit found things worth knowing before you trust its output.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | WARN |
| L1 | Static analysis of the code | NA |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | PASS |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- none-observed
- Network
- none-observed
- Shell
- none-observed
- Dependencies
- pinned
- Secrets in source
- none-found
Findings (5)
.claude/skills/glimpse-changes
.claude/skills/improve-codebase-architecture
.claude/skills/mintlify
.claude/skills/turborepo
.claude/skills/writing-tests
Gates applied: no_behavioural_pass.
eeed53996b1dfull audit observations/trust-audit/skill/saffron-health__get-pr-reviews.json · Report an issue / request a re-scanAudit history
Every audit this skill has had.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-13 | eeed53996b1d | CAUTION | B | 89 | first audit |
Questions
What does the Get Pr Reviews skill do?
The AI toolkit for building reliable browser automations
Is Get Pr Reviews safe to install?
With care. The audit graded it B (89/100) and found 5 things worth knowing before you trust this skill, listed below with the exact line each was found on.
What can Get Pr Reviews access on my machine?
The audit observed no filesystem, network or shell use at all in its source.
How current is this page?
The grade is for one exact copy of the source (eeed53996b1d), read on 2026-10-13. The repository is watched, and a new audit runs when it changes — this is the first audit.