diff --git a/.github/workflows/pages.yml b/.github/workflows/pages.yml new file mode 100644 index 0000000..32286b3 --- /dev/null +++ b/.github/workflows/pages.yml @@ -0,0 +1,37 @@ +name: Publish product website + +on: + workflow_dispatch: + +permissions: + contents: read + pages: write + id-token: write + +concurrency: + group: pages + cancel-in-progress: false + +jobs: + build: + runs-on: ubuntu-latest + steps: + - name: Check out the selected source + uses: actions/checkout@v6 + - name: Configure GitHub Pages + uses: actions/configure-pages@v5 + - name: Upload the static website + uses: actions/upload-pages-artifact@v4 + with: + path: website + + deploy: + needs: build + runs-on: ubuntu-latest + environment: + name: github-pages + url: ${{ steps.deployment.outputs.page_url }} + steps: + - name: Deploy to GitHub Pages + id: deployment + uses: actions/deploy-pages@v4 diff --git a/website/index.html b/website/index.html new file mode 100644 index 0000000..2340cb7 --- /dev/null +++ b/website/index.html @@ -0,0 +1,95 @@ + + + + + + + + Reflex Control — Check the work before the agent finishes + + + + +
+
+ Reflex Control + +
+
+
+
+
Open source · Rust · Early prototype
+

Check the work before your AI agent calls it done.

+

Reflex Control checks protected-file writes and selected risky commands, then runs your configured tests when a coding agent attempts to finish.

+
Explore the project ↗Get v0.5.0
+
For independent developers and small teams.
Start with the documented Claude Code integration.
+
+
+
reflex / agent-turnILLUSTRATION
+
+

01 / agent requests a write
target: .env
→ write denied · protected path

+

02 / agent attempts to finish
run: configured project tests
→ a test fails · return failure to agent

+

agent revises the change
run: configured project tests
→ tests pass · turn may finish

+ A bounded retry loop. Human review still matters. +
+
Simplified example. See the live test record for observed behavior.
+
+
+
v0.5.0 · Published October 2, 2026MIT · Source availableClaude Code · Live test documented
+
+
01 / THE WORKFLOW

Three checkpoints.
One agent turn.

Hooks add checks around an agent's actions. Your configured rules determine what is protected and which tests count as verification.

+
+ + + +
+
+

Set boundaries for file changes.

Check supported file writes against configured protected paths. The default policy includes environment files, private keys, and secrets directories.

+
Illustrative checkpointWrite → .envDenied → protected path
+
+

Hook behavior varies by agent. These checks are not a sandbox. Read the adapter coverage and limitations ↗

+
+
+
02 / WHAT EXISTS TODAY

A released prototype.
A trace you can inspect.

A documented Claude Code session includes blocked writes to protected paths and a failing test followed by a correction and a passing test. The record describes the setup, observations, and untested paths.

This is early research evidence, not production validation. Other adapters have different levels of verification.

+ +
+
+
Planned · Not yet available

Give test failures
more useful context.

The next proposed feature is optional Claude API analysis of user-selected code diffs and test-failure evidence: explain the issue and suggest concrete next steps.

The existing Claude Code hooks and this proposed API feature are separate. Content selection and sensitive-data handling need implementation and evaluation before user trials.

+
  1. 01Build the optional Claude API integration
  2. 02Evaluate explanations against known failures
  3. 03Track request cost and cap retries
  4. 04Try the workflow with initial volunteers
+
+
+
03 / THE BUILDER

Built by Murat.

I build Rust-based tools for AI-assisted software development, with a focus on verification, bounded execution, and inspectable results. Reflex Control sits alongside my agent runtime and repository intelligence experiments.

+
GitHub / rustfuture ↗Rust Agent Runtime ↗Repository Intelligence ↗
+
+
+ +
+ + +