Commit Graph

5 Commits

Author SHA1 Message Date
Tom Boucher
13c635f795 feat(security): improve prompt injection scanner — invisible Unicode, encoding obfuscation, structural validation, entropy analysis (#1839)
* fix(tests): allowlist execute-phase.md in prompt-injection scan

execute-phase.md grew to ~51K chars after the code-review gate step
was added in #1630, tripping the 50K size heuristic in the injection
scanner. The limit is calibrated for user-supplied input — trusted
workflow source files that legitimately exceed it are allowlisted
individually, following the same pattern as discuss-phase.md.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* feat(security): improve prompt injection scanner with 4 detection layers (#1838)

- Layer 1: Unicode tag block U+E0000–U+E007F detection in strict mode (2025 supply-chain attack vector)
- Layer 2: Character-spacing obfuscation, delimiter injection (<system>/<assistant>/<user>/<human>), and long hex sequence patterns
- Layer 3: validatePromptStructure() — validates XML tag structure of agent/workflow files against known-valid tag set
- Layer 4: scanEntropyAnomalies() — Shannon entropy analysis flagging high-entropy paragraphs (>5.5 bits/char)

All layers implemented TDD (RED→GREEN): 31 new tests written first, verified failing, then implemented.
Full suite: 2559 tests, 0 failures. security.cjs: 99.6% stmt coverage.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

---------

Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-04-05 20:22:52 -04:00
Tom Boucher
3078279a9a feat(planner): add reachability_check step (#1606)
* feat(planner): add reachability_check step to prevent unreachable code

Closes #1495

* fix: trim gsd-planner.md below 50000-char limit after rebase

The reachability_check addition pushed the file to 50,275 chars when
merged with the assign_waves additions from #1600. Condense both sections
while preserving all logic; file is now 49,859 chars.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix(security): normalize CRLF before prompt-stuffing length check

On Windows, git checks out files with CRLF line endings. JavaScript's
String.length counts \r characters, so a 49,859-byte file measures as
51,126 chars on Windows — falsely tripping the 50,000-char security
scanner. Normalize CRLF → LF before measuring in security.cjs and in
the reachability-check test.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* fix: trim gsd-planner.md to stay under 50000-char limit after merge with main

After rebasing onto main (which added mcp_tool_usage block), combined content
reached 50031 chars. Remove suggested log format from assign_waves rule to
bring file to 49972 chars, well under the 50000-char security scanner limit.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

---------

Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-04-03 13:00:41 -04:00
Tom Boucher
e03a9edd44 fix: replace invalid \Z regex anchor and remove redundant pattern
The original PR (#1337) used \Z in a JavaScript regex, which is a
Perl/Python/Ruby anchor — JavaScript interprets it as a literal match
for the character 'Z', silently truncating expected text containing
that letter. Replace with a two-pass approach: try next-key lookahead
first, fall back to greedy match to end-of-string.

Also remove the redundant `to=all:` pattern in sanitizeForDisplay()
since it is a subset of the existing `to=[^:\s]+:` pattern.

Add regression tests proving the Z-truncation bug and verifying
expected blocks at end-of-section parse correctly.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-03-23 22:19:49 -04:00
3metaJun
aaaa8e96fe Harden verify-work checkpoint rendering 2026-03-23 14:05:33 -04:00
Tom Boucher
62db008570 security: add prompt injection guards, path traversal prevention, and input validation
Defense-in-depth security hardening for a codebase where markdown files become
LLM system prompts. Adds centralized security module, PreToolUse hook for
injection detection, and CI-ready codebase scan.

New files:
- security.cjs: path traversal prevention, prompt injection scanner/sanitizer,
  safe JSON parsing, field name validation, shell arg validation
- gsd-prompt-guard.js: PreToolUse hook scans .planning/ writes for injection
- security.test.cjs: 62 unit tests for all security functions
- prompt-injection-scan.test.cjs: CI scan of all agent/workflow/command files

Hardened code paths:
- readTextArgOrFile: path traversal guard (--prd, --text-file)
- cmdStateUpdate/Patch: field name validation prevents regex injection
- cmdCommit: sanitizeForPrompt strips invisible chars from commit messages
- gsd-tools --fields: safeJsonParse wraps unprotected JSON.parse
- cmdFrontmatterGet/Set: null byte rejection
- cmdVerifyPathExists: null byte rejection
- install.js: registers prompt guard hook, updates uninstaller

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-03-20 11:38:26 -04:00