Merge pull request #1751 from entireio/fix/redact-layer-vocab-and-pins

fix(redact): reconcile layer-count vocabulary and pin provider-token boundaries

Author: gtrrz-victor

cmd/entire/cli/checkpoint/checkpoint_test.go

@@ -4695,7 +4695,7 @@ func TestCheckpointSummary_HasReview(t *testing.T) {
 // Summary.Intent and ReviewPrompt that previously bypassed redaction because
 // the dispatcher only matched .jsonl. The PR 1236 fix extended the JSON-aware
 // branch to .json. We assert via a low-entropy AWS-key shaped secret (catches
-// the 7-layer pipeline) so the test stays deterministic without the OPF binary.
+// the regex-only pipeline) so the test stays deterministic without the OPF binary.
 func TestRedactBlobBytes_JSONMetadata(t *testing.T) {
 	t.Parallel()
 

cmd/entire/cli/checkpoint/persistent.go

@@ -448,7 +448,7 @@ func (s *treeWriter) applyTranscriptBackfill(ctx context.Context, opts UpdateOpt
 		}
 	}
 
-	// Replace prompts with 7-layer-redacted content.
+	// Replace prompts with regex-only-redacted content.
 	if len(opts.Prompts) > 0 {
 		promptContent := RedactedJoinedPrompts(opts.Prompts)
 		blobHash, err := CreateBlobFromContent(s.repo, []byte(promptContent))
@@ -715,7 +715,7 @@ func (s *treeWriter) writeSessionToSubdirectory(ctx context.Context, opts WriteO
 	}
 	filePaths.AssetsManifest = manifestPath
 
-	// Write prompts via the 7-layer pipeline. OPF runs only in the
+	// Write prompts via the regex-only pipeline. OPF runs only in the
 	// pre-push rewrite path (manual_commit_opf_rewrite.go).
 	if len(opts.Prompts) > 0 {
 		promptContent := RedactedJoinedPrompts(opts.Prompts)
@@ -2254,12 +2254,13 @@ func (s *treeWriter) copyMetadataDir(ctx context.Context, metadataDir, sessionDi
 			return fmt.Errorf("path traversal detected: %s", relPath)
 		}
 
-		// Create blob from file with 7-layer secrets redaction.
-		// Post-commit emits 7-layer-only blobs; the pre-push rewrite
+		// Create blob from file with regex-only secrets redaction (the
+		// eight always-on/opt-in layers).
+		// Post-commit emits regex-only blobs; the pre-push rewrite
 		// (strategy/manual_commit_opf_rewrite.go) walks the resulting
 		// tree, re-redacts these blobs with OPF when enabled, and
-		// rewrites entire/checkpoints/v1 into 8-layer commits before
-		// they leave the local machine.
+		// rewrites entire/checkpoints/v1 into OPF-applied (9-layer)
+		// commits before they leave the local machine.
 		blobHash, mode, err := createRedactedBlobFromFile(ctx, s.repo, path, relPath)
 		if err != nil {
 			return fmt.Errorf("failed to create blob for %s: %w", path, err)
@@ -2281,12 +2282,13 @@ func (s *treeWriter) copyMetadataDir(ctx context.Context, metadataDir, sessionDi
 	return nil
 }
 
-// createRedactedBlobFromFile reads a file, applies the 7-layer redaction
-// pipeline, and creates a git blob. Used by committed-checkpoint writes
-// at post-commit time. The OpenAI Privacy Filter is intentionally NOT
-// run here — OPF lives in the pre-push rewrite path
-// (strategy/manual_commit_opf_rewrite.go), which re-redacts the 7-layer
-// blobs into 8-layer commits before they leave the local machine.
+// createRedactedBlobFromFile reads a file, applies the regex-only redaction
+// pipeline (the eight always-on/opt-in layers), and creates a git blob. Used
+// by committed-checkpoint writes at post-commit time. The OpenAI Privacy
+// Filter is intentionally NOT run here — OPF lives in the pre-push rewrite
+// path (strategy/manual_commit_opf_rewrite.go), which re-redacts the
+// regex-only blobs into OPF-applied (9-layer) commits before they leave the
+// local machine.
 // JSONL files get JSONL-aware redaction; all other files get plain byte redaction.
 func createRedactedBlobFromFile(ctx context.Context, repo *git.Repository, filePath, treePath string) (plumbing.Hash, filemode.FileMode, error) {
 	info, err := os.Stat(filePath)
@@ -2328,8 +2330,8 @@ func createRedactedBlobFromFile(ctx context.Context, repo *git.Repository, fileP
 // JSON-shaped files (.jsonl or .json) get JSON-aware redaction (falling
 // back to plain bytes on parse failure so regex/credential layers
 // still apply); other files get plain byte redaction. When
-// usePrivacyFilter is true the full 8-layer pipeline (including OPF)
-// runs; otherwise the 7-layer pipeline.
+// usePrivacyFilter is true the full 9-layer pipeline (the eight regex
+// layers plus OPF) runs; otherwise just the eight regex layers.
 //
 // .json is handled alongside .jsonl because checkpoint metadata files
 // (metadata.json, per-session metadata.json) carry free-form fields

cmd/entire/cli/checkpoint/persistent_opf_trailer_test.go

@@ -18,12 +18,12 @@ import (
 
 // TestWriteCommitted_DoesNotEmitOPFAppliedTrailer is the regression guard
 // for the architectural promise: standard post-commit condensation writes
-// 7-layer-only blobs and MUST NOT mark them with the Entire-OPF-Applied
+// regex-only blobs and MUST NOT mark them with the Entire-OPF-Applied
 // trailer. The trailer is emitted exclusively by the pre-push rewrite
 // path; if a future change accidentally added it to the standard writer,
 // the pre-push rewrite would skip those commits (HasOPFApplied true →
-// reparent-only, no actual OPF run) and ship 7-layer content as if it
-// were 8-layer. This test pins down that contract.
+// reparent-only, no actual OPF run) and ship regex-only content as if it
+// were OPF-applied. This test pins down that contract.
 func TestWriteCommitted_DoesNotEmitOPFAppliedTrailer(t *testing.T) {
 	t.Parallel()
 

cmd/entire/cli/checkpoint/prompts.go

@@ -23,8 +23,9 @@ func SplitPromptContent(content string) []string {
 	return prompts
 }
 
-// RedactedJoinedPrompts joins prompts and runs the 7-layer redaction
-// pipeline. OPF runs exclusively in the pre-push rewrite (not here),
+// RedactedJoinedPrompts joins prompts and runs the regex-only redaction
+// pipeline (the eight always-on/opt-in layers). OPF runs exclusively in
+// the pre-push rewrite (not here),
 // so the writer's hot path stays predictable. Exported so alternate
 // persistent backends produce identically-redacted prompt blobs.
 func RedactedJoinedPrompts(prompts []string) string {

cmd/entire/cli/checkpoint/prompts_test.go

@@ -28,8 +28,9 @@ func TestSplitPromptContent_EmptyContent(t *testing.T) {
 }
 
 // TestRedactedJoinedPrompts_AppliesSafetyNet verifies the helper joins
-// prompts with the canonical separator and runs them through the 7-layer
-// pipeline. OPF runs only in the pre-push rewrite path, never here.
+// prompts with the canonical separator and runs them through the
+// regex-only pipeline. OPF runs only in the pre-push rewrite path, never
+// here.
 func TestRedactedJoinedPrompts_AppliesSafetyNet(t *testing.T) {
 	t.Parallel()
 	got := RedactedJoinedPrompts([]string{"hello", "world"})

cmd/entire/cli/settings/settings.go

@@ -284,7 +284,7 @@ type OPFSettings struct {
 
 	// PromptDefault controls whether the pre-push hook asks the user
 	// before running OPF. "" (default) and "ask" both surface the
-	// interactive prompt; "never" skips OPF and pushes 7-layer content;
+	// interactive prompt; "never" skips OPF and pushes regex-only content;
 	// "always" runs without asking. ENTIRE_OPF=yes|no on the push
 	// invocation overrides this setting per-push.
 	PromptDefault string `json:"prompt_default,omitempty"`

cmd/entire/cli/strategy/common.go

@@ -442,7 +442,7 @@ func EnsureRedactionConfigured() {
 			})
 		}
 
-		// OpenAI Privacy Filter (opt-in 8th layer).
+		// OpenAI Privacy Filter (opt-in 9th layer).
 		if s.Redaction != nil && s.Redaction.OpenAIPrivacyFilter != nil {
 			opf := s.Redaction.OpenAIPrivacyFilter
 			redact.ConfigurePrivacyFilter(redact.OPFConfig{

cmd/entire/cli/strategy/manual_commit_condensation.go

@@ -105,11 +105,13 @@ type condenseOpts struct {
 	allAgentFiles    map[string]struct{} // Union of all sessions' FilesTouched for cross-session exclusion (nil = single-session)
 }
 
-// redactSessionJSONLBytes runs the 7-layer redaction pipeline over a
-// session transcript at post-commit condensation. OPF is intentionally
-// NOT included here — it runs exclusively in the pre-push rewrite path
+// redactSessionJSONLBytes runs the regex-only redaction pipeline (the
+// eight always-on/opt-in layers) over a session transcript at
+// post-commit condensation. OPF is intentionally NOT included here —
+// it runs exclusively in the pre-push rewrite path
 // (strategy/manual_commit_opf_rewrite.go), which re-redacts the
-// 7-layer blobs and produces 8-layer commits before the push.
+// regex-only blobs and produces OPF-applied (9-layer) commits before
+// the push.
 //
 // Exposed as a var so tests can inject deterministic success/error
 // returns. The signature still takes a context so the var can be
@@ -355,7 +357,7 @@ func (s *ManualCommitStrategy) CondenseSession(ctx context.Context, repo *git.Re
 		summary = generateSummary(ctx, redactedTranscript, sessionData.FilesTouched, state)
 	}
 
-	// Post-commit emits 7-layer-only blobs. OPF runs later in the
+	// Post-commit emits regex-only blobs. OPF runs later in the
 	// pre-push rewrite path, never here.
 	skillEvents := mergeSkillEvents(state.SkillEvents, withSkillEventTurnID(sessionData.SkillEvents, state.TurnID))
 

cmd/entire/cli/strategy/manual_commit_hooks.go

@@ -2842,9 +2842,9 @@ func (s *ManualCommitStrategy) finalizeAllTurnCheckpoints(ctx context.Context, s
 	// (attribution, files touched, prompts). Hooks run without user interaction
 	// so there is no retry path — preserving partial metadata is better than
 	// losing everything. Persisting an unredacted transcript would be worse.
-	// Run the 7-layer pipeline over the transcript — OPF runs later in
-	// the pre-push rewrite path, which re-redacts these 7-layer blobs
-	// and produces 8-layer commits before the push goes out.
+	// Run the regex-only pipeline over the transcript — OPF runs later in
+	// the pre-push rewrite path, which re-redacts these regex-only blobs
+	// and produces OPF-applied (9-layer) commits before the push goes out.
 	// Externalize inline images BEFORE redaction, mirroring CondenseSession, so the
 	// finalized (authoritative, full-session) transcript keeps its placeholders and
 	// matching assets instead of re-inlining what condensation lifted out. Opt-in;
@@ -2895,7 +2895,7 @@ func (s *ManualCommitStrategy) finalizeAllTurnCheckpoints(ctx context.Context, s
 		redactedTranscript = redact.RedactedBytes{}
 	}
 
-	// Post-commit emits 7-layer-only blobs; the writer joins + redacts
+	// Post-commit emits regex-only blobs; the writer joins + redacts
 	// via checkpoint.redactedJoinedPrompts. OPF runs later, once per
 	// push, in the pre-push rewrite path.
 	stores, err := checkpoint.Open(ctx, repo, checkpoint.OpenOptions{})

cmd/entire/cli/strategy/manual_commit_opf_prompt.go

@@ -21,8 +21,8 @@ import (
 type OPFDecision int
 
 const (
-	OPFRun   OPFDecision = iota // run the rewrite, push 8-layer
-	OPFSkip                     // skip the rewrite, push 7-layer
+	OPFRun   OPFDecision = iota // run the rewrite, push OPF-applied (9-layer)
+	OPFSkip                     // skip the rewrite, push regex-only (8-layer)
 	OPFAbort                    // cancel the push entirely (Ctrl-C / non-TTY abort)
 )