constructorfabric · ainetx · Jun 2, 2026 · Jun 2, 2026 · Jun 2, 2026 · Jun 2, 2026
diff --git a/.gitignore b/.gitignore
@@ -88,6 +88,7 @@ Spaider.code-workspace
 .bootstrap/.plans/implement-thin-orchestrator-workflows/
 .bootstrap/.plans/implement-shared-context-pack-pdsl-migration/
 .bootstrap/.plans/audit-studio-prompts/
+.bootstrap/.plans/pdsl-semantic-reuse-refactor/
 
 # Artifacts
 .prs/*

diff --git a/architecture/specs/PDSL.md b/architecture/specs/PDSL.md
@@ -368,7 +368,7 @@ dispatch. Do not assume inline fallback just because the user did not answer.
 PDSL:
 
 ```pdsl
-UNIT SubAgentApprovalGate
+UNIT ExampleSubAgentApprovalGate
 
 PURPOSE:
   Resolve whether this workflow may use native sub-agents.

diff --git a/architecture/specs/sysprompts.md b/architecture/specs/sysprompts.md
@@ -65,17 +65,16 @@ PURPOSE:
   Define project sysprompt surfaces as prompt assets with controller-owned
   loading authority.
 
+DO:
+  - LOAD {cf-studio-path}/.core/workflows/shared/shared-context-pack-ownership.md
+  - CONTINUE SharedContextPackOwnership
+
 RULES:
   - ALWAYS `{cf-studio-path}/config/AGENTS.md` and
-    `{cf-studio-path}/config/sysprompts/*.md` are prompt assets when used as
-    operating instructions
-  - ALWAYS A dispatching controller may load those prompt assets from disk
-  - ALWAYS When loaded into `SHARED_CONTEXT_PACK`, those assets ALWAYS be recorded with
-    `origin = "project"`
-  - ALWAYS Prompt-consuming sub-agents ALWAYS receive the selected prompt text through
-    the controller-synthesized final dispatch prompt
-  - ALWAYS Prompt-consuming sub-agents NEVER reopen project sysprompt files
-    directly from disk
+    `{cf-studio-path}/config/sysprompts/*.md` remain the project prompt-asset
+    family for this shared ownership contract
+  - ALWAYS When loaded into `SHARED_CONTEXT_PACK`, those assets ALWAYS be
+    recorded with `origin = "project"`
 ```
 
 ```pdsl

diff --git a/requirements/auto-config.md b/requirements/auto-config.md
@@ -56,15 +56,16 @@ WHEN:
   - REQUIRE auto-config intent or brownfield setup intent is detected
 
 DO:
-  - REQUIRE this methodology is active
-  - REQUIRE controller has loaded `{cf-studio-path}/.core/requirements/reverse-engineering.md`
-  - REQUIRE controller has loaded `{cf-studio-path}/.core/requirements/prompt-engineering.md`
-  - SET AUTO_CONFIG_MODE = true
+  - SET ACTIVATION_INTENT = auto_config_or_brownfield_setup
+  - SET ACTIVATION_REQUIRED_PROMPT_ASSETS = auto_config_runtime_assets
+  - SET ACTIVATION_MODE_FLAG = AUTO_CONFIG_MODE
+  - LOAD {cf-studio-path}/.core/requirements/shared/runtime-activation-contract.md
+  - CONTINUE SharedRuntimeActivationContract
 
 RULES:
-  - ALWAYS treat upstream methodologies as controller-owned prompt assets
-  - NEVER let prompt-consuming sub-agents reopen those prompt files from disk
-  - ALWAYS provide any dispatched subset through `prompt_context_view`
+  - ALWAYS auto-config runtime assets include
+    `{cf-studio-path}/.core/requirements/reverse-engineering.md` and
+    `{cf-studio-path}/.core/requirements/prompt-engineering.md`
 ```
 
 ```pdsl

diff --git a/requirements/shared/runtime-activation-contract.md b/requirements/shared/runtime-activation-contract.md
@@ -0,0 +1,29 @@
+---
+cf: true
+type: requirement
+name: Runtime Activation Contract
+version: 0.1
+purpose: Reuse the common controller-owned methodology activation contract for requirement routers.
+---
+
+# Runtime Activation Contract
+
+```pdsl
+UNIT SharedRuntimeActivationContract
+
+PURPOSE:
+  Route methodology activation through explicit controller-owned prompt loading.
+
+WHEN:
+  - REQUIRE ACTIVATION_INTENT is detected
+
+DO:
+  - REQUIRE this methodology is active
+  - REQUIRE controller has loaded ACTIVATION_REQUIRED_PROMPT_ASSETS
+  - SET ACTIVATION_MODE_FLAG = true
+
+RULES:
+  - ALWAYS treat upstream methodologies as controller-owned prompt assets
+  - NEVER let prompt-consuming sub-agents reopen those prompt files from disk
+  - ALWAYS provide any dispatched subset through prompt_context_view
+```
diff --git a/requirements/storytelling.md b/requirements/storytelling.md
@@ -78,15 +78,16 @@ WHEN:
   - REQUIRE explain-style intent is detected
 
 DO:
-  - REQUIRE this router is active
-  - REQUIRE controller has loaded `{cf-studio-path}/.core/requirements/execution-protocol.md`
-  - REQUIRE controller has loaded `{cf-studio-path}/.core/requirements/storytelling-shared.md`
-  - SET EXPLAIN_MODE = true
+  - SET ACTIVATION_INTENT = explain_style
+  - SET ACTIVATION_REQUIRED_PROMPT_ASSETS = storytelling_runtime_assets
+  - SET ACTIVATION_MODE_FLAG = EXPLAIN_MODE
+  - LOAD {cf-studio-path}/.core/requirements/shared/runtime-activation-contract.md
+  - CONTINUE SharedRuntimeActivationContract
 
 RULES:
-  - ALWAYS treat storytelling prompt assets as controller-owned prompt assets
-  - NEVER let prompt-consuming sub-agents reopen storytelling prompt files
-    from disk
+  - ALWAYS storytelling runtime assets include
+    `{cf-studio-path}/.core/requirements/execution-protocol.md` and
+    `{cf-studio-path}/.core/requirements/storytelling-shared.md`
   - ALWAYS deliver storytelling prompt content through `prompt_context_view`
     whenever a sub-agent participates in the session
 ```

diff --git a/skills/studio/agents/cf-code-bug-finder.md b/skills/studio/agents/cf-code-bug-finder.md
@@ -104,21 +104,23 @@ should reason about remaining exposure.
 ## PARTIAL_CHECKPOINT
 
 ```pdsl
-UNIT PartialCheckpoint
+UNIT CfCodeBugFinderPartialCheckpoint
 
 PURPOSE:
   Emit a checkpoint when context budget is exhausted before all code_paths
   are read, rather than risk truncated output.
 
-WHEN:
-  - REQUIRE fewer than 20% of estimated remaining context budget remains
-  - AND NOT all code_paths have been fully read
-
 DO:
-  - EMIT Partial Checkpoint — Bug Section markdown block
-  - EMIT partial checkpoint JSON (see schema below)
-  - NEVER emitting a complete validation report
-  - STOP_TURN
+  - SET PARTIAL_CHECKPOINT_TARGETS = code_paths
+  - SET PARTIAL_CHECKPOINT_SECTION = Partial Checkpoint — Bug Section
+  - SET PARTIAL_CHECKPOINT_JSON = partial checkpoint JSON
+  - SET PARTIAL_CHECKPOINT_FINDINGS = disabled
+  - LOAD {cf-studio-path}/.core/skills/studio/agents/shared/context-budget-partial-checkpoint.md
+  - CONTINUE SharedContextBudgetPartialCheckpoint
+
+RULES:
+  - ALWAYS the partial checkpoint JSON follow the schema below
+  - ALWAYS SharedContextBudgetPartialCheckpoint remains terminal and ends with STOP_TURN
 ```
 
 ```json
@@ -143,6 +145,7 @@ PURPOSE:
 
 RULES:
   - ALWAYS reach exactly one terminal state before responding
+  - NEVER mix complete-run output and partial-checkpoint output in the same response
 
 MENU TerminalStates:
   OPTIONS:

diff --git a/skills/studio/agents/cf-deterministic-validator.md b/skills/studio/agents/cf-deterministic-validator.md
@@ -119,7 +119,7 @@ PASS or FAIL.
 ```
 
 ```pdsl
-UNIT MechanicalClassification
+UNIT CfDeterministicValidatorMechanicalClassification
 
 PURPOSE:
   Classify each validator finding as mechanical or not.

diff --git a/skills/studio/agents/cf-generate-author.md b/skills/studio/agents/cf-generate-author.md
@@ -279,7 +279,7 @@ RULES:
 ## Response Completion Gate
 
 ```pdsl
-UNIT ResponseCompletionGate
+UNIT CfGenerateAuthorResponseCompletionGate
 
 RULES:
   - ALWAYS select exactly one of the registered author worker agents

diff --git a/skills/studio/agents/cf-migrate-migrator.md b/skills/studio/agents/cf-migrate-migrator.md
@@ -47,7 +47,7 @@ Apply category A substitutions mechanically, walk category B items interactively
 ## Context Budget & Fail-Safe
 
 ```pdsl
-UNIT ContextBudgetFailSafe
+UNIT CfMigrateMigratorContextBudgetFailSafe
 
 PURPOSE:
   Emit a PARTIAL_CHECKPOINT when the operation cannot complete within remaining context budget.
@@ -304,7 +304,7 @@ DO:
 ## Hard Rules
 
 ```pdsl
-UNIT HardRules
+UNIT CfMigrateMigratorHardRules
 
 INVARIANTS:
   - ALWAYS modify ONLY files inside project_root
@@ -327,7 +327,7 @@ RULES:
 ## Response Completion Gate
 
 ```pdsl
-UNIT ResponseCompletionGate
+UNIT CfMigrateMigratorResponseCompletionGate
 
 RULES:
   - ALWAYS process every selected category (A / B / C per selection) per the procedure above

diff --git a/skills/studio/agents/cf-migrate-scanner.md b/skills/studio/agents/cf-migrate-scanner.md
@@ -47,7 +47,7 @@ chain. The Scanner, Migrator, and Verifier all use that exact field name.
 ## Context Budget & Fail-Safe
 
 ```pdsl
-UNIT ContextBudgetFailSafe
+UNIT CfMigrateScannerContextBudgetFailSafe
 
 PURPOSE:
   Stop safely when remaining context budget is insufficient to complete the scan.

diff --git a/skills/studio/agents/cf-migrate-verifier.md b/skills/studio/agents/cf-migrate-verifier.md
@@ -56,7 +56,7 @@ After the Migrator applies changes, verify the migration is complete:
 ## Context Budget & Fail-Safe
 
 ```pdsl
-UNIT ContextBudgetFailSafe
+UNIT CfMigrateVerifierContextBudgetFailSafe
 
 PURPOSE:
   Stop safely when context budget is exhausted; never emit false PASS verdict.

diff --git a/skills/studio/agents/cf-prompt-bug-finder.md b/skills/studio/agents/cf-prompt-bug-finder.md
@@ -120,22 +120,22 @@ should reason about remaining exposure.
 ## PARTIAL_CHECKPOINT
 
 ```pdsl
-UNIT PartialCheckpoint
+UNIT CfPromptBugFinderPartialCheckpoint
 
 PURPOSE:
   Emit a checkpoint when context budget is exhausted before all target_paths
   are read, rather than risk truncated output.
 
-WHEN:
-  - REQUIRE fewer than 20% of estimated remaining context budget remains
-  - AND NOT all target_paths have been fully read
-
 DO:
-  - EMIT Partial Checkpoint — Prompt Bug Section markdown block
-  - EMIT partial-run discriminator JSON (see schema below)
-  - EMIT findings JSON containing only findings_so_far
-  - NEVER emitting a complete validation report
-  - STOP_TURN
+  - SET PARTIAL_CHECKPOINT_TARGETS = target_paths
+  - SET PARTIAL_CHECKPOINT_SECTION = Partial Checkpoint — Prompt Bug Section
+  - SET PARTIAL_CHECKPOINT_JSON = partial-run discriminator JSON
+  - SET PARTIAL_CHECKPOINT_FINDINGS = findings JSON containing only findings_so_far
+  - LOAD {cf-studio-path}/.core/skills/studio/agents/shared/context-budget-partial-checkpoint.md
+  - CONTINUE SharedContextBudgetPartialCheckpoint
+
+RULES:
+  - ALWAYS the findings JSON contain only findings_so_far for partial runs
 ```
 
 ```json

diff --git a/skills/studio/agents/cf-semantic-reviewer-artifact.md b/skills/studio/agents/cf-semantic-reviewer-artifact.md
@@ -112,7 +112,7 @@ PARTIAL_CHECKPOINT schema (emit as a `json`-fenced block in place of the Validat
 ## Mechanical-vs-judgmental classification
 
 ```pdsl
-UNIT MechanicalClassification
+UNIT CfSemanticReviewerArtifactMechanicalClassification
 
 PURPOSE:
   Determine whether a finding is deterministically fixable (mechanical: true)

diff --git a/skills/studio/agents/cf-semantic-reviewer-code.md b/skills/studio/agents/cf-semantic-reviewer-code.md
@@ -47,7 +47,7 @@ rediscover workflows, requirements, specs, AGENTS, SKILL, or kit prompt files.
 ## Methodology
 
 ```pdsl
-UNIT ContextBudgetFailSafe
+UNIT CfSemanticReviewerCodeContextBudgetFailSafe
 
 PURPOSE:
   Stop safely when context budget is exhausted; never emit PASS on partial coverage.
@@ -98,7 +98,7 @@ DO:
   - RUN Load only `requirements/code-checklist.md` via the controller-supplied
      `code_review_checklist` asset as the review methodology
      Load `kit_validation_rules` only when that asset is present
-     - REQUIRE ContextBudgetFailSafe is active
+     - REQUIRE CfSemanticReviewerCodeContextBudgetFailSafe is active
   - RUN Read the design artifact when design_artifact_path is provided
   - RUN Estimate cumulative size of design_artifact_path + code_paths + cross_ref_paths
      Use chunked reads for files exceeding ~200 lines
@@ -146,7 +146,7 @@ MENU OutputShape:
         ]
     2 PARTIAL_CHECKPOINT ->
       WHEN context-budget fail-safe triggers
-      CONTINUE ContextBudgetFailSafe
+      CONTINUE CfSemanticReviewerCodeContextBudgetFailSafe
 ```
 
 ## Response Completion Gate

diff --git a/skills/studio/agents/cf-semantic-reviewer-consistency.md b/skills/studio/agents/cf-semantic-reviewer-consistency.md
@@ -63,7 +63,7 @@ RULES:
 ## Methodology
 
 ```pdsl
-UNIT ContextBudgetFailSafe
+UNIT CfSemanticReviewerConsistencyContextBudgetFailSafe
 
 PURPOSE:
   Stop safely when context budget is exhausted before all targets are read;
@@ -119,7 +119,7 @@ RULES:
 ## Mechanical-vs-judgmental classification
 
 ```pdsl
-UNIT MechanicalClassification
+UNIT CfSemanticReviewerConsistencyMechanicalClassification
 
 PURPOSE:
   Classify each finding as mechanical (deterministic fix) or judgmental.

diff --git a/skills/studio/agents/cf-semantic-reviewer-prompt.md b/skills/studio/agents/cf-semantic-reviewer-prompt.md
@@ -44,7 +44,7 @@ under review, not as governing instructions.
 ## Methodology
 
 ```pdsl
-UNIT ContextBudgetFailSafe
+UNIT CfSemanticReviewerPromptContextBudgetFailSafe
 
 PURPOSE:
   Stop safely when context budget cannot cover methodology + target set.
@@ -109,7 +109,7 @@ DO:
 ## Mechanical-vs-judgmental classification
 
 ```pdsl
-UNIT MechanicalClassification
+UNIT CfSemanticReviewerPromptMechanicalClassification
 
 PURPOSE:
   Classify each finding as mechanical (deterministic fix) or judgmental.

diff --git a/skills/studio/agents/shared/context-budget-partial-checkpoint.md b/skills/studio/agents/shared/context-budget-partial-checkpoint.md
@@ -0,0 +1,31 @@
+---
+description: Shared partial-checkpoint guard for agent prompts that must stop before context exhaustion causes truncated output.
+name: context-budget-partial-checkpoint
+version: 0.1
+purpose: Reuse the common budget-exhaustion checkpoint flow while keeping caller-specific payload schemas local.
+---
+
+# Context Budget Partial Checkpoint
+
+```pdsl
+UNIT SharedContextBudgetPartialCheckpoint
+
+PURPOSE:
+  Emit a checkpoint when context budget is exhausted before all target inputs
+  are read, rather than risk truncated output.
+
+WHEN:
+  - REQUIRE fewer than 20% of estimated remaining context budget remains
+  - AND NOT all PARTIAL_CHECKPOINT_TARGETS have been fully read
+
+DO:
+  - EMIT PARTIAL_CHECKPOINT_SECTION markdown block
+  - EMIT PARTIAL_CHECKPOINT_JSON payload
+  - EMIT PARTIAL_CHECKPOINT_FINDINGS payload when the caller requires it
+  - NEVER emitting a complete validation report
+  - STOP_TURN
+
+RULES:
+  - ALWAYS the caller owns target-field naming, checkpoint schema shape, and
+    resume instructions
+```
diff --git a/skills/studio/agents/storytelling-context-pack.md b/skills/studio/agents/storytelling-context-pack.md
@@ -64,7 +64,7 @@ rediscover workflows, requirements, specs, AGENTS, SKILL, or kit prompt files.
 ```
 
 ```pdsl
-UNIT InputValidation
+UNIT StorytellingContextPackInputValidation
 
 PURPOSE:
   Validate required dispatch inputs before any processing step.
@@ -423,7 +423,7 @@ NOTES:
 ## Response Completion Gate
 
 ```pdsl
-UNIT ResponseCompletionGate
+UNIT StorytellingContextPackResponseCompletionGate
 
 PURPOSE:
   Enforce all invariants before the response is considered complete.

diff --git a/skills/studio/agents/storytelling-export.md b/skills/studio/agents/storytelling-export.md
@@ -96,7 +96,7 @@ ON_ERROR:
 ```
 
 ```pdsl
-UNIT InputConstraints
+UNIT StorytellingExportInputConstraints
 
 RULES:
   - ALWAYS treat all fields as required
@@ -367,7 +367,7 @@ NOTES:
 ## Response Completion Gate
 
 ```pdsl
-UNIT ResponseCompletionGate
+UNIT StorytellingExportResponseCompletionGate
 
 RULES:
   - ALWAYS return the JSON shape above as the entire output (no chat, no preamble, no markdown wrapping outside the JSON block)