docs: add written procedures to Ollama and PR review guides#315
docs: add written procedures to Ollama and PR review guides#315rachaelrenk wants to merge 11 commits into
Conversation
Add thorough step-by-step written procedures to two high-traffic guides that previously had video content but thin written coverage. how-to-set-up-ollama.mdx: - Add Prerequisites section with VRAM/RAM requirements table - Add platform-specific install instructions (macOS, Linux, Windows) - Add proper model pull/run commands with real model names - Add model types glossary (updated formatting) - Add model performance comparison section with timing tips - Add OpenAI-compatible API integration with Python and Node.js examples - Add Modelfile customization section with worked example - Add Productivity tips and Next steps sections how-to-review-prs-like-a-senior-dev.mdx: - Rewrite opening with clearer framing (structured review vs. AI summaries) - Add Prerequisites section - Add 'Why structured review matters' conceptual section - Add step-by-step procedure: get diff, attach context, use prompt - Keep existing structured prompt template (preserved verbatim) - Add section-by-section explanation of the prompt output - Add Tips for iterating with example follow-up prompts - Add Next steps with related guides Co-Authored-By: Oz <oz-agent@warp.dev>
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
|
I'm starting a first review of this pull request. You can view the conversation on Warp. I completed the review and no human review was requested for this pull request. Comment Powered by Oz |
![oz-for-oss[bot]](https://avatars.githubusercontent.com/in/3450139?s=60&v=4){kind=small}
There was a problem hiding this comment.
Overview
This PR expands two guides with written, step-by-step procedures for PR review and Ollama setup. The structure is useful, but the Ollama benchmarking section documents a command that will not produce the timing stats it tells readers to compare.
Concerns
- The
ollama runbenchmark example needs--verbosebefore it can show token timing stats. - The runnable GitHub CLI examples use angle-bracket placeholders instead of the docs style's ALL_CAPS placeholder values.
Verdict
Found: 1 critical, 0 important, 1 suggestions
Request changes
Comment /oz-review on this pull request to retrigger a review (up to 3 times on the same pull request).
Powered by Oz
After cross-checking against official Ollama docs: - Fix macOS minimum version: macOS 12+ → macOS 14 Sonoma+ - Fix macOS install: replace non-official brew path with official curl script (same script as Linux); add DMG as manual alternative - Fix Windows install: add official PowerShell one-liner (irm https://ollama.com/install.ps1 | iex) alongside manual installer - Fix CLI command: ollama list → ollama ls (official canonical command) - Fix base URL: add trailing slash to http://localhost:11434/v1/ per official OpenAI compat docs - Fix REST API curl example: /api/generate → /api/chat with messages format - Remove undocumented preloading tip (ollama run <model> "") - Add --verbose flag for timing stats instead of claiming they print by default - Mention native ollama-python/ollama-js SDKs alongside OpenAI compat path - Fix gh pr view --web comment (was incorrectly 'copy the diff URL') Co-Authored-By: Oz <oz-agent@warp.dev>
Co-authored-by: oz-for-oss[bot] <277970191+oz-for-oss[bot]@users.noreply.github.com>
…-a-senior-dev.mdx Co-authored-by: oz-for-oss[bot] <277970191+oz-for-oss[bot]@users.noreply.github.com>
- Remove 'rule of thumb' colloquialism; replace with 'as a general estimate' - Remove em dash from Apple Silicon sentence per suggestion (use period) - Consolidate benchmark example to single --verbose command so timing stats are actually visible (critical fix from oz-for-oss review) - Replace <PR-number> angle-bracket placeholders with PR_NUMBER (ALL_CAPS per docs style guide) in both the code block and inline reference Co-Authored-By: Oz <oz-agent@warp.dev>
1. SDK package names: ollama-python/ollama-js (GitHub repo names) -> 'pip install ollama' and 'npm i ollama' (actual package names) 2. Remove 'Quantization' from model types table — it is not a label displayed in ollama.com/library; it is a property of model tag variants (e.g. :q4_0). Update column header to 'Label' to match the actual UI. Quantized model paragraph below the table is kept. 3. Phi-3 Mini clarified as 3.8B in the 3B row (ollama library: phi3:3.8b) 4. 30-34B row corrected: Codestral is 22B, not 30-34B; Mixtral is 46.7B total params (MoE). Row label changed to 22-34B. 5. Linux systemd claim softened: the curl installer does not automatically enable a service on all distros; replaced with explicit systemctl commands to start and enable Ollama. 6. ollama ls described as showing 'last-used dates' -> 'download dates' (the MODIFIED column reflects when a model was pulled, not used) Co-Authored-By: Oz <oz-agent@warp.dev>
|
|
||
| <VideoEmbed url="https://youtu.be/NVwqQyphlAw?si=zcMR1ZHt-xnIS_ME" title="Reviewing pull requests with Warp Agent video" /> | ||
|
|
||
| ## Prerequisites |
There was a problem hiding this comment.
I think we should rethink the framing here. Right now this reads like “paste this giant template and get a senior-engineer-quality review,” which feels a bit heavy handed.
A stronger guide would focus on the actual workflow:
- Give the agent the PR goal and relevant context
- Ask it to inspect the diff and surrounding code
- Prioritize correctness, regressions, security, and missing tests
- Verify any findings before leaving review comments
- Use follow-ups to investigate specific concerns
The structured output can still be useful, but it should support the review process rather than be the main point of the guide.
There was a problem hiding this comment.
a few specific changes:
- drop most of the emojis
- ask the agent to inspect surrounding code
- add an explicit verification step (link to computer use)
- use the GH pR directly wherever possible:
gh pr diffonly gives the patch.gh pr view, the checked-out-branch, tests, etc, would provide more context
generally, i think we can simplify this instead of recommending such a specific step by step guide to code review
| import { Steps } from '@astrojs/starlight/components'; | ||
|
|
||
| Learn how to prompt Warp’s AI to review pull requests like an experienced engineer — focusing on structure, red flags, and clarity. | ||
| AI can summarize a pull request in seconds — but a summary isn't a review. This guide shows you how to prompt Warp's agent to produce structured, prioritized feedback that mirrors how a senior engineer would approach a large PR: start with risk, surface blockers, and give a clear merge recommendation. Plan on about 10 minutes to complete. |
There was a problem hiding this comment.
This opening feels too promotional and makes a pretty strong claim that the output will “mirror how a senior engineer” reviews a PR. I’d make this more grounded... something like:
“Warp’s agent can help you understand a large pull request, identify risky changes, and investigate potential issues before you leave review comments. This guide walks through how to give the agent enough context, ask for focused feedback, and verify its findings.”
| --- | ||
|
|
||
| <Steps> | ||
| ### 1. 🚨 Risk Assessment |
There was a problem hiding this comment.
i worry that this template would feel like ai slop to a senior dev and they wouldn't read it
| - No account required — Ollama is free and open source | ||
|
|
||
| ### 2. Run your first model | ||
| ## 1. Install Ollama |
There was a problem hiding this comment.
@oliviajohnsto - do we want to do some co-marketing with ollama here perhaps?
|
|
||
| ## 5. Integrate Ollama into your app | ||
|
|
||
| Ollama exposes an OpenAI-compatible REST API at `http://localhost:11434/v1/`. You can drop it in wherever you're currently using the OpenAI API — just update three values: |
There was a problem hiding this comment.
you may want to confirm with an agent because i think this needs to be a publicly available URL for it to work with Warp's custom endpoints (which you should mention). i.e. in our docs somewhere else, we mention that you can expose this via an ngrok tunnel
| Explore related guides and features: | ||
|
|
||
| This adds a structured system prompt for that task — ready to use instantly. | ||
| - [Set up Claude Code](/guides/external-tools/how-to-set-up-claude-code/) or [Set up Codex CLI](/guides/external-tools/how-to-set-up-codex-cli/) — use cloud-based agents alongside your local Ollama setup |
There was a problem hiding this comment.
it should be to use Warp with this BYO inference, rather than set up claude code. we should push to Warp's agent here
- Replace over-claiming opening ('mirrors how a senior engineer') with
grounded copy focused on what the agent can actually help with
- Shift from template-first to workflow-first structure:
1. Give context (PR goal, diff, @file refs, run tests first)
2. Ask for focused feedback (concise, goal-oriented prompt)
3. Verify findings (run tests, read code yourself, computer use)
4. Follow-up prompts for specific concerns
- Demote structured output template to a 'Reference (optional)' section
- Strip most emojis from the template (kept [SECURITY] etc. prefix
convention which is scannable; removed colored-circle risk indicators
and emoji section headers)
- Remove the 'Formatting rules' section from the template (instructions
to the model folded into footer of the template)
- Expand context-gathering step: gh pr view for PR goal, @file refs for
surrounding code, running tests before prompting
- Add verification step with link to Computer use docs
- Remove 'Why structured review matters' section (no longer needed;
opening paragraph and workflow steps make the value clear)
- Update frontmatter description to match new framing
Addresses: 3562549711, 3562548536, 3562552949, 3562554524
Co-Authored-By: Oz <oz-agent@warp.dev>
2 participants
Summary
Add thorough step-by-step written procedures to two high-traffic guides that previously had video content but thin or missing written coverage.
Notion tracking item: https://app.notion.com/p/38143263616d81e58b9fc1522c554ac7
how-to-set-up-ollama.mdx(2,433 visitors / 3 months)The page previously had a video embed and narrative sections with no actual install commands or runnable code. Added:
ollama pull/ollama runcommands with real model names;ollama ls/ollama psreference--verboseflag for timing stats,ollama psfor memory usage- Compare model performance — non-interactive prompt mode,--verboseflag for timing stats,ollama psfor memory usage- **Comparnative Ollama SDKsModelfilewalkthrough with a worked example (system prompt + parameters)how-to-review-prs-like-a-senior-dev.mdx(225 visitors / 3 months)The page previously had The page previously had The page previously had The page previously had The page previously had The page previously had The pageed:
Generated with Oz
Co-Authored-By: Oz oz-agent@warp.dev