diff --git a/CHANGELOG.md b/CHANGELOG.md
index bb5d15e..2b34430 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -24,6 +24,11 @@ All notable changes to this project will be documented here.
   titled `Conduct` as a workaround for GitHub's lack of a general
   private-contact channel. Closes #25.
 
+### Fixed
+- Pipe (`|`) and backslash (`\`) in table cell values are now escaped in the
+  `markdown` and `notion` output formats, preventing column-structure corruption
+  in rendered tables. Closes #16.
+
 ### Changed
 - Renamed `.github/workflows/test.yml` → `ci.yml` and the workflow
   `name:` from `Tests` to `CI` for cross-repo consistency. README
diff --git a/docs/superpowers/specs/2026-04-11-table-formatter-escaping-design.md b/docs/superpowers/specs/2026-04-11-table-formatter-escaping-design.md
new file mode 100644
index 0000000..5c31802
--- /dev/null
+++ b/docs/superpowers/specs/2026-04-11-table-formatter-escaping-design.md
@@ -0,0 +1,84 @@
+# Table Formatter Escaping -- Design Spec
+
+**Date:** 2026-04-11
+**Issues:** #16, #17, #18 (skipping #15 -- potential contributor)
+**Scope:** Fix pipe-character escaping in Markdown/Notion and Jira/Confluence formatters; add escaping test matrix for all output formats.
+
+## Problem
+
+`_format_markdown` and `_format_jira` in `parser.py` interpolate raw cell values into pipe-delimited table syntax. A cell containing `|` breaks the table structure in every downstream renderer. Backslash `\` also needs escaping since it is the escape character in both formats.
+
+Zero tests cover special characters in any formatter.
+
+## Approach
+
+Per-formatter escape helpers (Approach A). Each formatter gets its own escape function. The logic is identical today (`\` -> `\\`, then `|` -> `\|`), but separate helpers keep each formatter self-contained and extensible for format-specific concerns later.
+
+## PR 1: Fix #16 -- Markdown/Notion pipe escaping
+
+### Changes in `src/mcp_clipboard/parser.py`
+
+Add helper:
+
+```python
+def _escape_md_cell(cell: str) -> str:
+    """Escape characters that break GFM pipe tables."""
+    return cell.replace("\\", "\\\\").replace("|", "\\|")
+```
+
+Modify `_format_markdown()`:
+- Apply `_escape_md_cell()` to every cell value in header and data rows
+- Compute column widths from the *escaped* values so alignment is correct
+- Apply escaping *before* `ljust()` padding
+
+Notion routes through `_format_markdown` -- no additional changes needed.
+
+### Tests
+
+Add tests in `tests/test_parser.py` (alongside existing formatter tests):
+- Cell containing `|` -- assert column count preserved, pipe escaped in output
+- Cell containing `||` -- same
+- Cell containing `\` -- assert escaped to `\\`
+- Cell containing `\|` -- assert escaped to `\\|` (no ambiguity)
+- Leading/trailing `|` in a cell
+
+## PR 2: Fix #17 -- Jira/Confluence pipe escaping
+
+### Changes in `src/mcp_clipboard/parser.py`
+
+Add helper:
+
+```python
+def _escape_jira_cell(cell: str) -> str:
+    """Escape characters that break Jira/Confluence wiki markup tables."""
+    return cell.replace("\\", "\\\\").replace("|", "\\|")
+```
+
+Modify `_format_jira()`:
+- Apply `_escape_jira_cell()` to every cell value in header and data rows
+
+Confluence routes through `_format_jira` -- no additional changes needed.
+
+### Tests
+
+Add tests in `tests/test_parser.py`:
+- Cell containing `|` -- assert column count preserved, pipe escaped
+- Cell containing `||` -- assert no accidental header promotion
+- Cell containing `\` -- assert escaped
+- Leading/trailing `|` in a cell
+
+## PR 3: Fix #18 -- Escaping test matrix
+
+### New file: `tests/test_escaping.py`
+
+Parametrized test matrix covering every `OutputFormat` with special characters.
+
+### Merge strategy
+
+PR 3 rebases onto main after PRs 1 and 2 are merged. Markdown/Notion/Jira/Confluence pipe tests are passing assertions (not xfails).
+
+## Out of scope
+
+- #15: HTML XSS (`_format_html` cell escaping) -- potential contributor
+- #19: Slack `*` and backtick escaping -- medium priority, separate PR
+- #20: `detect_content_type` false positives -- unrelated
diff --git a/src/mcp_clipboard/parser.py b/src/mcp_clipboard/parser.py
index 28929b4..970462c 100644
--- a/src/mcp_clipboard/parser.py
+++ b/src/mcp_clipboard/parser.py
@@ -356,6 +356,11 @@ def format_table(rows: list[list[str]], fmt: OutputFormat = "markdown") -> str:
         return _format_markdown(rows)
 
 
+def _escape_md_cell(cell: str) -> str:
+    """Escape characters that break GFM pipe tables."""
+    return cell.replace("\\", "\\\\").replace("|", "\\|")
+
+
 def _format_markdown(rows: list[list[str]]) -> str:
     """Render rows as a GitHub-flavored Markdown table."""
     if not rows:
@@ -365,9 +370,11 @@ def _format_markdown(rows: list[list[str]]) -> str:
     max_cols = max(len(row) for row in rows)
     normalized = [row + [""] * (max_cols - len(row)) for row in rows]
 
-    # Calculate column widths
+    # Escape cells, then calculate widths from escaped values
+    escaped = [[_escape_md_cell(cell) for cell in row] for row in normalized]
+
     widths = [
-        max(len(normalized[r][c]) for r in range(len(normalized)))
+        max(len(escaped[r][c]) for r in range(len(escaped)))
         for c in range(max_cols)
     ]
     # Minimum width of 3 for the separator
@@ -376,7 +383,7 @@ def _format_markdown(rows: list[list[str]]) -> str:
     lines: list[str] = []
     # Header
     header = "| " + " | ".join(
-        normalized[0][c].ljust(widths[c]) for c in range(max_cols)
+        escaped[0][c].ljust(widths[c]) for c in range(max_cols)
     ) + " |"
     lines.append(header)
 
@@ -385,7 +392,7 @@ def _format_markdown(rows: list[list[str]]) -> str:
     lines.append(sep)
 
     # Data rows
-    for row in normalized[1:]:
+    for row in escaped[1:]:
         line = "| " + " | ".join(
             row[c].ljust(widths[c]) for c in range(max_cols)
         ) + " |"
diff --git a/tests/test_parser.py b/tests/test_parser.py
index f1f910b..c9d2128 100644
--- a/tests/test_parser.py
+++ b/tests/test_parser.py
@@ -1,5 +1,7 @@
 """Tests for the HTML/TSV parser, formatters, and content detection."""
 
+import re
+
 from mcp_clipboard.parser import (
     detect_content_type,
     extract_html_text,
@@ -464,3 +466,82 @@ def test_format_destination_ragged_rows():
     for fmt in ("slack", "jira", "confluence", "html"):
         result = format_table(rows, fmt)
         assert result  # non-empty
+
+
+# ---------------------------------------------------------------------------
+# Markdown/Notion escaping (#16)
+# ---------------------------------------------------------------------------
+
+
+def test_format_markdown_escapes_pipe_in_data():
+    """Pipe in a data cell must not break column structure."""
+    rows = [["Cmd", "Desc"], ["cat | grep", "filter"]]
+    result = format_table(rows, "markdown")
+    lines = result.strip().split("\n")
+    # Header + separator + 1 data row
+    assert len(lines) == 3
+    # Every line must have the same number of unescaped pipe delimiters
+    # (escaped pipes \| do not count as delimiters)
+    # 2-column table: | col1 | col2 | = 3 unescaped pipes per line
+    for line in lines:
+        # Count unescaped pipes: pipes not preceded by backslash
+        delimiters = len(re.findall(r"(?<!\\)\|", line))
+        assert delimiters == 3, f"Expected 3 pipe delimiters, got {delimiters} in: {line}"
+    # The escaped pipe must appear in the output
+    assert r"cat \| grep" in result
+
+
+def test_format_markdown_escapes_pipe_in_header():
+    """Pipe in a header cell must not break column structure."""
+    rows = [["A|B", "C"], ["1", "2"]]
+    result = format_table(rows, "markdown")
+    lines = result.strip().split("\n")
+    assert len(lines) == 3
+    assert r"A\|B" in result
+
+
+def test_format_markdown_escapes_double_pipe():
+    """Double pipe || must not create extra columns."""
+    rows = [["X", "Y"], ["a||b", "c"]]
+    result = format_table(rows, "markdown")
+    for line in result.strip().split("\n"):
+        delimiters = len(re.findall(r"(?<!\\)\|", line))
+        assert delimiters == 3  # 2-column table: | col1 | col2 | = 3 unescaped pipes
+
+
+def test_format_markdown_escapes_backslash():
+    r"""Backslash must be escaped so \| in source doesn't become a bare pipe."""
+    rows = [["Val"], [r"a\b"]]
+    result = format_table(rows, "markdown")
+    assert r"a\\b" in result
+
+
+def test_format_markdown_escapes_backslash_pipe():
+    r"""Source \| must become \\| (escaped backslash + escaped pipe)."""
+    rows = [["Val"], [r"a\|b"]]
+    result = format_table(rows, "markdown")
+    assert r"a\\\|b" in result
+
+
+def test_format_markdown_leading_trailing_pipe():
+    """Leading/trailing pipes in cell values must be escaped."""
+    rows = [["Col"], ["|leading"], ["trailing|"], ["|both|"]]
+    result = format_table(rows, "markdown")
+    lines = result.strip().split("\n")
+    # header + sep + 3 data rows = 5 lines
+    assert len(lines) == 5
+    for line in lines:
+        delimiters = len(re.findall(r"(?<!\\)\|", line))
+        assert delimiters == 2  # single column: | cell |
+    assert r"\|leading" in result
+    assert r"trailing\|" in result
+    assert r"\|both\|" in result
+
+
+def test_format_notion_escapes_pipe():
+    """Notion uses _format_markdown, so escaping must apply there too."""
+    rows = [["A", "B"], ["x|y", "z"]]
+    md_result = format_table(rows, "markdown")
+    notion_result = format_table(rows, "notion")
+    assert md_result == notion_result
+    assert r"x\|y" in notion_result