diff --git a/.github/ISSUE_TEMPLATE/bug-report.md b/.github/ISSUE_TEMPLATE/bug-report.md
new file mode 100644
index 0000000..e52777b
--- /dev/null
+++ b/.github/ISSUE_TEMPLATE/bug-report.md
@@ -0,0 +1,31 @@
+---
+name: Bug report
+description: Report a reproducible benchmark or CLI issue using sanitized evidence
+title: "[bug] "
+labels: []
+assignees: []
+---
+
+> Public issue: do not paste credentials, private URLs/IPs, provider account details, raw production logs, customer data, or infrastructure details.
+
+## Observed behavior
+
+Describe what `proxybench` returned or did.
+
+## Expected behavior
+
+Describe the deterministic behavior you expected.
+
+## Minimal synthetic reproducer
+
+Provide the smallest sanitized JSONL example that reproduces the issue.
+
+## Environment
+
+- Python version:
+- `proxybench` version/commit:
+- OS/runtime, if relevant:
+
+## Additional context
+
+Include only non-sensitive details needed to understand the bug.
diff --git a/.github/pull_request_template.md b/.github/pull_request_template.md
new file mode 100644
index 0000000..81097fa
--- /dev/null
+++ b/.github/pull_request_template.md
@@ -0,0 +1,16 @@
+## Summary
+
+What changed and why?
+
+## Evidence
+
+What metric, test, or reproducible case supports the change?
+
+## Safety / scope check
+
+- [ ] No credentials, private endpoints/IPs, provider account details, customer data, or production captures are included.
+- [ ] Missing-data behavior remains explicit.
+- [ ] Descriptive comparisons are not presented as causal proof.
+- [ ] Public/private implementation boundaries are preserved.
+- [ ] Tests were added or updated when behavior changed.
+- [ ] Local tests and repository hygiene checks pass.
diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
new file mode 100644
index 0000000..0bcc5a2
--- /dev/null
+++ b/CONTRIBUTING.md
@@ -0,0 +1,46 @@
+# Contributing
+
+Thanks for considering a contribution to `proxybench`.
+
+## Scope
+
+Good contributions improve the public measurement baseline without weakening its data-minimization or evidence-first behavior.
+
+Useful changes include:
+
+- additional deterministic summary metrics;
+- clearer comparison semantics;
+- explicit missing-data coverage;
+- regression tests for edge cases;
+- CLI usability improvements;
+- packaging, CI, security, or documentation improvements.
+
+Private routing/scoring logic, provider selection, credentials, production infrastructure, private benchmark data, and customer/workload identifiers do not belong in this repository.
+
+## Development
+
+```bash
+python -m pip install .
+python -m unittest discover -s tests -v
+python scripts/public_hygiene_check.py
+```
+
+Keep runtime dependencies at zero unless there is a strong, documented reason to change that constraint.
+
+## Pull requests
+
+A good pull request should:
+
+1. state the metric or operator problem precisely;
+2. include tests for behavior changes;
+3. keep input/output semantics deterministic;
+4. report missing-data coverage rather than silently imputing values;
+5. avoid presenting descriptive comparisons as causal proof;
+6. keep fixtures synthetic and non-sensitive;
+7. keep CI green across the supported Python matrix.
+
+## Sensitive information
+
+Do not post credentials, private endpoints/IPs, provider account details, production logs, customer data, or private infrastructure details.
+
+See [`SECURITY.md`](SECURITY.md) for the repository security boundary.
diff --git a/README.md b/README.md
index 0627f01..a0911b7 100644
--- a/README.md
+++ b/README.md
@@ -1,11 +1,24 @@
 # proxybench
 
+> Evidence-first benchmarking for proxy and web-retrieval workloads.
+
 **Measure usable results, not proxy count.**
 
-`proxybench` is a local, deterministic benchmark utility for proxy and web-retrieval workloads. It compares sanitized request outcomes using metrics that matter to operators: usable success rate, requests per usable result, rotations per usable result, latency, and cost per usable result.
+`proxybench` is a local, deterministic Python utility for comparing sanitized retrieval outcomes using metrics that matter to operators: usable success rate, requests per usable result, rotations per usable result, latency, and cost per usable result.
+
+**Status:** alpha · Python 3.10+ · zero runtime dependencies · local-only · no telemetry
 
 Built by **PN Labs**.
 
+## PN Labs reliability toolkit
+
+| Project | Purpose |
+| --- | --- |
+| [**proxy-outcome**](https://github.com/pnlabs-dev/proxy-outcome) | Classify what happened without over-attributing the cause |
+| **proxybench** | Compare usable-result efficiency between retrieval policies |
+
+The tools are intentionally separate: classify evidence first, then benchmark whether a policy actually improves outcomes.
+
 ## Why this exists
 
 Adding more proxies or rotating more often does not guarantee better retrieval outcomes. A benchmark should answer a narrower question:
@@ -124,6 +137,10 @@ python scripts/public_hygiene_check.py
 
 CI installs the package before testing, smoke-tests the installed CLI, and runs a non-echoing vendor-neutral public repository hygiene gate.
 
+## Contributing
+
+Small, measurable improvements are welcome. See [`CONTRIBUTING.md`](CONTRIBUTING.md) before opening a pull request.
+
 ## Security
 
 See [`SECURITY.md`](SECURITY.md). Do not put production logs, credentials, private endpoint information, personal/customer data, or private infrastructure details into public benchmark fixtures or issues.