claude-code-sessions-sanitizer 0.3.0__py3-none-any.whl

ccs_sanitize/__init__.py

@@ -0,0 +1,15 @@
+"""ccs_sanitize — scrub Claude Code session JSONL for safe publication.
+
+Design: see `.claude/specs/prd-sanitizer.md` in the repository.
+
+Bumping `__version__` is required for any change that alters the produced
+output bytes (new rule semantics, new secret patterns, changed serialization),
+because the value is recorded in every `.scrubbed` sidecar (PRD section 10) and
+downstream consumers gate on it. See CHANGELOG.md for the bump policy and
+README.md ("Stability and the determinism contract") for the public
+statement of what each bump level promises.
+"""
+
+__version__ = "0.3.0"
+
+__all__ = ["__version__"]

ccs_sanitize/_templates/ccs-sanitize.example.yaml

@@ -0,0 +1,66 @@
+# Sanitizer config template. This file ships in two roles:
+#   - `.ccs-sanitize.example.yaml` -- committed, schema-only reference.
+#   - `.ccs-sanitize.yaml`         -- gitignored, live config; edit in
+#                                     your own PII values here.
+# `ccs-sanitize --init` writes both files from this same template; if
+# you set up by hand, copy this file to `.ccs-sanitize.yaml` and edit
+# the placeholders below. The live `.ccs-sanitize.yaml` is gitignored
+# because its match values are, by definition, the literal strings you
+# do not want committed to a public repo (real home dir, real email,
+# real name, etc.). See .gitignore for the convention rationale.
+#
+# The sidecar records only the BASENAME of the config under
+# `config_source:` (PRD section 10), so committed `.scrubbed` sidecars
+# stay safe regardless of what is in your local config. The leak surface
+# is the config file itself; gitignoring it closes the surface.
+#
+# Built-in secret patterns (PRD section 9, Tier 1 + Tier 2) are
+# non-weakenable and always apply on top of these rules.
+# Adding rules: the I-3 replacement-leak guard runs at load time, so a
+# misconfigured replacement (one matching another rule or a built-in
+# secret pattern) fails load with exit 3.
+version: 1
+
+paths:
+  # Home directory. /home/user is the canonical placeholder shipped by the
+  # CLI example so downstream consumers (sibling projects, fixture readers)
+  # treat it as the universal "developer's home directory."
+  - match: "/home/REAL_USERNAME"
+    replace: "/home/user"
+
+identifiers:
+  # Author email. Use the RFC-2606 reserved example.com TLD so a fixture
+  # cannot collide with a real address.
+  - match: "real.author@example.com"
+    replace: "user@example.com"
+
+  # Bare Unix username. Catches the project-dir encoding Claude Code uses
+  # (`-home-USERNAME-Documents-...` where slashes become dashes) plus
+  # `ls -l` owner/group columns and any chat content naming the user.
+  - match: "REAL_USERNAME"
+    replace: "user"
+
+  # GitHub handle. Catches github.com URLs in commit messages, tool
+  # output, and chat. Order this rule BEFORE the personal-name rule:
+  # a hyphenated handle like `first-mi-last` may overlap the surname.
+  - match: "real-github-handle"
+    replace: "real-user"
+
+  # Author personal name. Appears in git commit `Author:` lines, PRD
+  # `**Owner:**` fields, pyproject.toml `authors`, and agent prompts.
+  # The placeholder must differ from the match string -- a same-string
+  # mapping (e.g. "Real Name" -> "Real Name") fails the I-3 idempotency
+  # guard at load time (PRD section 10, config.py _check_replacement_leak).
+  - match: "Real Name"
+    replace: "Example Author"
+
+# options defaults are usually correct:
+#   scrub_git_branch: true   -- replace `gitBranch` field values with
+#                               `feature/example` placeholder
+#   remap_uuids: false       -- leave session/agent/parent UUIDs as-is so
+#                               the parent-subagent graph stays readable
+#                               in committed fixtures
+#
+# extra_secret_patterns: []  -- adds project-specific secret detectors on
+#                               top of the built-in Tier 1 + Tier 2
+#                               patterns; never removes them.