docs(bf-25hf0): add verification note for grep-corpus baseline

- Document acceptance criteria status (PASS/WARN items)
- Record corpus metrics and verification steps
- Note grep implementation blocked on 7.8.x beads

notes/bf-25hf0.md

@@ -0,0 +1,48 @@
+# bf-25hf0: grep-corpus baseline metrics
+
+## Task Completed
+
+Captured grep-corpus baseline metrics to `benches/baselines/grep-corpus.json`.
+
+## What Was Done
+
+1. **Validated corpus exists**: Confirmed 1,000 PDF files in `tests/fixtures/grep-corpus/corpus/` totaling ~6.6 MB
+2. **Measured corpus metrics**: 
+   - File count: 1,000
+   - Total size: 6,870,672 bytes (~6.6 MB)
+3. **Recorded git metadata**:
+   - Commit SHA: 77eeaecc
+   - Timestamp: 2026-07-06T16:06:55+00:00
+4. **Created baseline JSON**: `benches/baselines/grep-corpus.json` with available metrics
+
+## Acceptance Criteria Status
+
+- ✅ **PASS**: `benches/baselines/grep-corpus.json` exists
+- ✅ **PASS**: JSON contains required fields (commit_sha, timestamp, corpus_size, bytes_total, notes)
+- ✅ **PASS**: JSON is valid and parseable (verified with `python3 -m json.tool`)
+- ✅ **PASS**: Corpus validation passes (1,000 files confirmed)
+- ✅ **PASS**: Timestamp is current (2026-07-06T16:06:55+00:00)
+- ✅ **PASS**: Commit hash matches HEAD (77eeaecc)
+- ⚠️ **WARN**: Grep subcommand not yet implemented (blocked on 7.8.x beads), timing/throughput metrics deferred
+
+## Notes
+
+The grep functionality is not yet implemented in pdftract CLI. The benchmark code at `crates/pdftract-cli/benches/grep_1000.rs` is a TODO placeholder waiting for grep subcommand implementation (beads 7.8.x). 
+
+This baseline captures the corpus state before implementation. Once grep is implemented, this baseline should be updated with actual timing metrics (throughput_mb_per_sec, files_per_sec, total_runtime_sec).
+
+## Artifacts
+
+- `benches/baselines/grep-corpus.json` - baseline metrics file
+- Commit: a9a7eb0d (on main branch)
+
+## Verification
+
+```bash
+# Validate JSON
+cat benches/baselines/grep-corpus.json | python3 -m json.tool
+
+# Verify corpus
+ls -1 tests/fixtures/grep-corpus/corpus/*.pdf | wc -l  # 1000 files
+du -sb tests/fixtures/grep-corpus/corpus/ | awk '{print $1}'  # 6870672 bytes
+```