A pull request review agent for Claude Code, Codex and Cursor
PR Reviewer reviews one pull request's diff for bugs and for breaches of the workspace's decisions, lessons and preferences, citing the line and the constraint for each finding. A run costs 1,200 credits, charged only when the run succeeds.
Or use it on slipstream.contextstream.io, signed in with your ContextStream account. Version 0.1.0 · Updated
What does PR Reviewer do?
Reviews a pull request's diff against the workspace's decisions, lessons and preferences, and returns findings that each cite a line of the diff and the constraint they rely on.
What it declines
A task like one of these is declined with a reason, and a declined run is free:
Changing, committing, approving on the host or merging anything: it returns findings for a person to act on.
Style nits a formatter or linter catches, unless a workspace preference asks for them.
Fixing failing CI (CI Fixer does that) or writing release notes (Release Notes does that).
Reviewing code that is not in the pull request's diff.
What does a real run of PR Reviewer look like?
The input, the result (shortened where it is long), how long it took, what it cost and where it ran. Live runs used public or demo material only; a run marked as from the eval suite is not a live run.
Live run on ContextStream
A review of a public, merged pull request
Version
0.1.0
Outcome
succeeded
Ran
2026-10-27 17:27 UTC
Took
51.3 s
Cost
1,200 credits
ContextStream's own workspace, reviewing pull request #221 of github.com/contextstream/mcp-server against that workspace's lessons for the repository.
Reviewed #221 "feat: MCP 1.0.34 shows which files are exact-search only": 3 findings (0 high, 3 medium, 0 low) in 9 files, 153 changed lines.
Verdict
comment: Reviewed #221 "feat: MCP 1.0.34 shows which files are exact-search only": 3 findings (0 high, 3 medium, 0 low) in 9 files, 153 changed lines.
Pull request
#221 feat: MCP 1.0.34 shows which files are exact-search only (9 files, 153 changed lines read)
Findings
Severity
Where
Finding
Suggestion
medium
crates/mcp-tools/src/domains/project.rs:2724
index_status clause only in text, not structured content
Include the formatted clause (or a structured `exact_search_only` summary) in the index_status structuredContent and assert on it in the tool-level test.
medium
crates/mcp-tools/src/domains/search.rs:7392
Exact-search note only in text, not structured content
Add the exact-search-only label to the search tool's structured result (e.g. a per-hit field or the structured text field) from the same `exact_search_only_label` source, and assert it in a test that inspects `structured_content`.
Checks: 4 of 4 passed
diff_read: Read 9 files of 10; 1 file not reviewed.
citations: Every finding cites a line of the diff and, when it relies on one, a constraint the review was given.
constraints: Checked against 12 decisions, 12 lessons and 0 preferences; 3 findings rely on one.
workspace_context: 0 context items from ContextStream for this job, about 480 tokens.
The lessons these findings rely on are from ContextStream’s own workspace, so their text and the findings’ explanations are withheld here.
1 more finding is withheld: it rests on an internal release lesson.
Task
0abc9a1e-33d3-4af1-aa27-b28f083d9d39
Charged
0: it ran on ContextStream’s own account, which is not billed; a customer pays the price shown
What does PR Reviewer cost?
PR Reviewer costs 1,200 ContextStream credits a run, charged only when the run succeeds; a declined, failed, cancelled or timed-out run costs nothing. Credits come with every plan: Starter is free with 10,000 a month (8 runs of PR Reviewer, if spent on nothing else), and Pro is $19 a month with 25,000.
Which tools does PR Reviewer work with?
Claude Code, Codex, Cursor and other MCP clients: ContextStream's MCP gives them a slipstream tool that finds, runs and follows PR Reviewer. It also runs from a terminal or an agent's shell with ss run review, and from the ContextStream web app.
Claude Code, Codex, Cursor and other MCP clients: connect ContextStream's MCP once (setup for each editor), then ask your agent to hand the job to PR Reviewer.
A terminal or an agent's shell:ss run review, installed with ContextStream's command line.
The web app: the workspace's specialist library, where each run shows its price before it starts.
What access does PR Reviewer need?
PR Reviewer gets a short-lived token for one workspace, limited to 6 scopes: Reads repos, Reads memory, Reads context, Marks cited context, Models via ContextStream and Can save a note. It works in one project, and only for that run.
Model agent. A program on ContextStream's model proxy, with a short-lived token that acts for you in one workspace. It is stopped after 4 min.
Scope
What it lets a run do
vcs.read
Read the workspace’s connected repositories: pull requests, commits, comparisons and tags.
memory.read
Search memory, and read decisions and lessons.
context.read
Read the run’s context pack from the workspace.
context.feedback
Mark the context items a result cited as helpful.
model.invoke
Call models through ContextStream’s model proxy, under the run’s credit cap.
memory.write.note
Save one note per run.
How do I turn PR Reviewer on?
An owner or admin of your ContextStream workspace turns PR Reviewer on in the specialist library of the web app. Specialist agents are opening workspace by workspace; if yours is not open yet, the library says so.
Sign in to ContextStream: the button below opens the specialist library on PR Reviewer.
An owner or admin switches PR Reviewer on for the workspace.
Pick the project it works in, and run it from the library, from your coding agent or with ss run.
How do I call PR Reviewer from Claude Code or Codex?
Ask your coding agent to hand the job to PR Reviewer, or call it yourself: ss run review in a terminal, or slipstream(action="run", agent="review", input={...}) over MCP. A priced run starts only once its price is accepted.
Connect ContextStream's MCP (it also writes each editor's connection):
Release Notes writes user-facing release notes for the changes between two refs of a connected repository, from its merged pull requests and commits, and saves them as a doc.
CI Fixer reproduces a failing test command in a public repository and returns the smallest change to the project's code that makes it pass, with the passing output.