@cleocode/skills 2026.9.21 → 2026.9.22

package/skills/ct-master-tac/bundled/protocols/architecture-decision.cant

@@ -1,90 +0,0 @@
----
-kind: protocol
-version: 1.0.0
-id: ADR
-title: "Architecture Decision Record Protocol"
-status: active
-type: conditional
-audience: "llm-agent, orchestrator"
-tags: "adr, architecture, decisions"
-skillRef: ct-adr-recorder
-lastUpdated: 2026-04-08
-enforcement: advisory
-stage: "architecture-decision"
-consult-when: "Recording a formal architecture decision after consensus reaches PROVEN verdict; superseding an existing ADR; any task carrying --protocol adr flag"
-input:
-  consensus_manifest_id: "string — ID of the originating consensus manifest (required)"
-  decision_title: "string — short title for the ADR"
-  options_evaluated: "string[] — options that were considered"
-output:
-  adr_id: "string — canonical ADR identifier (ADR-NNN)"
-  status: "proposed | accepted | superseded"
-  stored_at: "string — path to .cleo/adrs/<id>.md and decisions table row"
----
-
-# Architecture Decision Record (ADR) Protocol
-#
-# Provenance: @task T4798 (ADR-006 Implementation), @task T428 (gap patch)
-# Type: Conditional Protocol
-# Stage: RCADSD - A (ADR)
-# Max Active: 3 protocols (including base)
-#
-# Trigger Conditions
-#
-# This protocol activates when the task involves:
-#   Decision Recording: "decision", "adr", "architecture decision"
-#   Stage Transition: "after consensus", "begin adr"
-#   Formalization: "lock in decision", "formalize choice", "decide"
-#   Architectural Shift: "pivot", "new architecture", "supersede"
-#   Record Creation: "create adr", "write adr", "record decision"
-#
-# Explicit Override: --protocol adr flag on task creation.
-#
-# Requirements (RFC 2119)
-#
-# MUST:
-#   ADR-001: MUST be generated from an accepted Consensus report verdict
-#   ADR-002: MUST include a consensus_manifest_id linking to its originating consensus
-#   ADR-003: MUST require explicit HITL approval to transition from proposed to accepted
-#   ADR-004: MUST include Context, Options Evaluated, Decision, Rationale, and Consequences sections
-#   ADR-005: MUST trigger downstream invalidation if superseded
-#   ADR-006: MUST be stored in the canonical decisions SQLite table via Drizzle ORM
-#   ADR-007: MUST set agent_type: "decision" in manifest entry
-#   ADR-008: MUST block the Specification stage until the ADR status is accepted
-#
-# SHOULD:
-#   ADR-010: SHOULD document the exact data structures or schema changes required
-#   ADR-011: SHOULD explicitly list which existing ADRs are superseded, with rationale
-#   ADR-012: SHOULD flag known technical debt introduced by the decision
-#   ADR-013: SHOULD document rejected alternatives with rationale for rejection
-#
-# MAY:
-#   ADR-020: MAY include diagrams (Mermaid) illustrating the architectural shift
-#   ADR-021: MAY link to external prior art or research documents
-#   ADR-022: MAY reference related ADRs that are not superseded but contextually relevant
-#
-# Decision Status Lifecycle:
-#   proposed -> accepted -> superseded
-#                       \-> deprecated
-#
-# HITL Gate:
-#   1. Agent drafts the ADR based on consensus verdict
-#   2. Status is set to proposed
-#   3. Pipeline pauses (HANDOFF_REQUIRED - exit code 65)
-#   4. Human reviews the proposed ADR
-#   5. If approved, status transitions to accepted
-#   6. Only an accepted ADR unlocks the Specification stage
-#
-# Exit Codes:
-#   65: HANDOFF_REQUIRED - ADR drafted as proposed, awaiting HITL acceptance
-#   84: PROVENANCE_REQUIRED - Attempted to create ADR without linked Consensus report
-#   18: CASCADE_FAILED - Downstream work blocked because governing ADR was superseded
-#
-# Anti-Patterns:
-#   - Creating ADR without consensus (decisions lack evidence foundation)
-#   - Auto-accepting without HITL review (bypasses human oversight gate)
-#   - Omitting downstream impact section (future implementers unaware of cascade)
-#   - Superseding without updating specs (creates orphaned specifications)
-#   - Using ADR to define implementation requirements (that is Specification's role)
-#   - Storing ADR only as markdown without SQLite record (loses relational queries)
-#   - Skipping rejected alternatives (loses institutional knowledge)

package/skills/ct-master-tac/bundled/protocols/artifact-publish.cant

@@ -1,105 +0,0 @@
----
-kind: protocol
-version: 1.0.0
-id: ART
-title: "Artifact Publish Protocol"
-status: active
-type: cross-cutting
-audience: "llm-agent, orchestrator"
-tags: "artifact, publish, distribution"
-skillRef: ct-artifact-publisher
-lastUpdated: 2026-04-08
-enforcement: advisory
-stage: "release"
-consult-when: "Publishing artifacts to any registry (npm, cargo, docker, PyPI, GitHub releases); any task with --protocol artifact-publish; release pipeline artifact phase"
-input:
-  artifact_type: "npm-package | python-wheel | python-sdist | go-module | cargo-crate | ruby-gem | docker-image | github-release | generic-tarball"
-  version: "string — version string consistent with release config"
-  dry_run: "boolean — MUST be true on first invocation"
-output:
-  published_artifacts: "object[] — array of {type, name, registry_url, sha256, slsa_level}"
-  rollback_manifest: "string | null — path to rollback manifest if partial failure"
----
-
-# Artifact Publish Protocol
-#
-# Provenance: @task T428 (gap patch)
-# Type: Conditional Protocol
-# Max Active: 3 protocols (including base)
-#
-# Relationship to Release Protocol:
-#   This protocol orchestrates artifact building and publishing.
-#   Release orchestrates version bumping, tagging, and changelog.
-#   They compose: release triggers artifact-publish for distribution phase.
-#
-# Trigger Conditions
-#
-# This protocol activates when the task involves:
-#   Package Publish: "publish", "package", "distribute"
-#   Artifact Build: "artifact", "build artifact", "bundle"
-#   Container Push: "docker push", "container registry", "image publish"
-#   Language Package: "crate", "gem", "wheel", "sdist"
-#   Multi-Artifact: "publish all", "release artifacts", "multi-package"
-#
-# Explicit Override: --protocol artifact-publish flag on task creation.
-#
-# Requirements (RFC 2119)
-#
-# MUST:
-#   ARTP-001: MUST validate artifact configuration before build
-#   ARTP-002: MUST execute dry-run before any real publish
-#   ARTP-003: MUST follow handler interface contract: validate -> build -> publish
-#   ARTP-004: MUST generate SHA-256 checksums for all built artifacts
-#   ARTP-005: MUST record provenance metadata via record_release()
-#   ARTP-006: MUST use sequential execution for multi-artifact publish
-#   ARTP-007: MUST set agent_type: "artifact-publish" in manifest
-#   ARTP-008: MUST NOT store credentials in config, output, or manifest
-#   ARTP-009: MUST halt pipeline and attempt rollback on first publish failure
-#
-# SHOULD:
-#   ARTP-010: SHOULD verify registry reachability before publish
-#   ARTP-011: SHOULD validate version consistency between config and artifact metadata
-#   ARTP-012: SHOULD log all publish operations to audit trail
-#   ARTP-013: SHOULD verify build output exists and is non-empty before publish
-#
-# MAY:
-#   ARTP-020: MAY batch validation across all artifacts before starting builds
-#   ARTP-021: MAY generate SBOM alongside artifacts (delegate to provenance)
-#   ARTP-022: MAY sign artifacts using configured signing method (delegate to provenance)
-#
-# Artifact Lifecycle State Machine:
-#   configured -> validated -> built -> published
-#                    |            |         |
-#                  failed       failed    failed -> rollback
-#
-# Registered Handlers (9 types):
-#   npm-package, python-wheel, python-sdist, go-module,
-#   cargo-crate, ruby-gem, docker-image, github-release, generic-tarball
-#
-# Multi-Artifact Pipeline Phases:
-#   Phase 1: Pre-validate all artifacts
-#   Phase 2: Build sequential per artifact
-#   Phase 3: Publish with rollback on failure
-#
-# Credential Handling:
-#   Agents MUST NOT store, log, or embed credentials
-#   Resolution order: env var -> CI secret -> credential manager (future)
-#
-# Error Codes (85-89):
-#   85: E_ARTIFACT_TYPE_UNKNOWN
-#   86: E_ARTIFACT_VALIDATION_FAILED
-#   87: E_ARTIFACT_BUILD_FAILED
-#   88: E_ARTIFACT_PUBLISH_FAILED (rollback attempted)
-#   89: E_ARTIFACT_ROLLBACK_FAILED
-#
-# Anti-Patterns:
-#   - Publishing without dry-run first (irreversible registry state)
-#   - Storing credentials in config.json (committed to VCS)
-#   - Parallel multi-artifact publish (race conditions)
-#   - Skipping checksum generation (cannot verify integrity)
-#   - Publishing without version check (duplicate version errors)
-#   - Ignoring publish failures (inconsistent state)
-#   - Logging credential values (exposure in audit trail)
-#   - Building without validation (wastes time)
-#   - Manual rollback without recording (lost provenance chain)
-#   - Hardcoding registry URLs (breaks across environments)

package/skills/ct-master-tac/bundled/protocols/consensus.cant

@@ -1,85 +0,0 @@
----
-kind: protocol
-version: 1.0.0
-id: CONS
-title: "Consensus Protocol"
-status: active
-type: conditional
-audience: "llm-agent, orchestrator"
-tags: "consensus, voting, validation"
-skillRef: ct-consensus-voter
-lastUpdated: 2026-04-08
-enforcement: strict
-stage: "consensus"
-consult-when: "Reaching a decision that two or more agents must vote on; architecture choices, tool selection, policy decisions; any task carrying agent_type:analysis; phrases like 'reach consensus', 'vote on options', 'resolve the debate'"
-input:
-  topic: "string — the question or decision to vote on"
-  options: "string[] — candidate options agents vote on"
-  agents: "string[] — agent IDs participating in the vote"
-output:
-  verdict: "PROVEN | REFUTED | CONTESTED | INSUFFICIENT_EVIDENCE"
-  confidence: "number — weighted confidence score 0.0–1.0"
-  voting_matrix: "object — per-agent votes with rationale and confidence"
-  hitl_required: "boolean — true when threshold not reached"
----
-
-# Consensus Protocol
-#
-# Provenance: @task T3155, @epic T3147
-# Type: Conditional Protocol
-# Max Active: 3 protocols (including base)
-#
-# Trigger Conditions
-#
-# This protocol activates when the task involves:
-#   Decision Making: "vote", "decide", "choose", "select"
-#   Agreement: "consensus", "agree", "alignment"
-#   Conflict Resolution: "resolve", "dispute", "conflict"
-#   Validation: "validate claim", "verify assertion"
-#
-# Explicit Override: --protocol consensus flag on task creation.
-#
-# Requirements (RFC 2119)
-#
-# MUST:
-#   CONS-001: MUST use structured voting format
-#   CONS-002: MUST document rationale for each position
-#   CONS-003: MUST include confidence scores (0.0-1.0)
-#   CONS-004: MUST cite evidence supporting positions
-#   CONS-005: MUST flag conflicts with severity levels
-#   CONS-006: MUST escalate to HITL when threshold not reached
-#   CONS-007: MUST set agent_type: "analysis" in manifest
-#
-# SHOULD:
-#   CONS-010: SHOULD present multiple perspectives
-#   CONS-011: SHOULD identify hidden assumptions
-#   CONS-012: SHOULD document rejected alternatives
-#   CONS-013: SHOULD include uncertainty notes for low confidence
-#
-# MAY:
-#   CONS-020: MAY propose compromise positions
-#   CONS-021: MAY defer non-critical decisions
-#   CONS-022: MAY request additional research
-#
-# Verdict Thresholds:
-#   PROVEN: 3/5 agents OR 50%+ weighted confidence, reproducible evidence
-#   REFUTED: Counter-evidence invalidates, counter-proof exists
-#   CONTESTED: 3/5 split after 2 challenge rounds, document both sides
-#   INSUFFICIENT_EVIDENCE: Cannot reach verdict, request investigation
-#
-# HITL Escalation:
-#   - Contested verdict (3/5 split): Present conflict to user
-#   - Critical severity conflict: Immediate escalation
-#   - Insufficient evidence: Request user guidance
-#   - Unanimous suspicious consensus: Verify with user
-#
-# Exit Codes:
-#   EXIT_PROTOCOL_CONSENSUS (61) - Consensus protocol violation
-#   EXIT_PROTOCOL_GENERIC (67) - Generic protocol error
-#
-# Anti-Patterns:
-#   - Accepting unanimous consensus without scrutiny (may indicate groupthink)
-#   - Skipping evidence citations (decisions lack foundation)
-#   - Binary voting without confidence (loses nuance)
-#   - Ignoring minority positions (may miss valid concerns)
-#   - Premature escalation (wastes human attention)

package/skills/ct-master-tac/bundled/protocols/contribution.cant

@@ -1,92 +0,0 @@
----
-kind: protocol
-version: 1.0.0
-id: CONT
-title: "Contribution Protocol"
-status: active
-type: cross-cutting
-audience: "llm-agent, orchestrator"
-tags: "contribution, commit, pr"
-skillRef: ct-contribution
-lastUpdated: 2026-04-08
-enforcement: strict
-stage: "cross-cutting"
-consult-when: "Submitting work via git commit or PR; multi-agent consensus contribution; task carries 'contribution protocol', 'submit contribution', or 'consensus workflow'; any cross-cutting attribution recording"
-input:
-  task_id: "string — task being contributed"
-  commit_message: "string — conventional commit message"
-  contributors: "string[] — agent or human contributor IDs"
-output:
-  commit_sha: "string — resulting commit SHA"
-  pr_url: "string | null — pull request URL if opened"
-  attribution_recorded: "boolean — true when contribution row persisted"
----
-
-# Contribution Protocol
-#
-# Provenance: @task T3155, @epic T3147
-# Type: Cross-Cutting Protocol
-# Applies To: All RCASD-IVTR+C stages
-# Max Active: 3 protocols (including base)
-#
-# Cross-Cutting Nature: This protocol applies across ALL stages of RCASD-IVTR+C.
-# Unlike stage-specific protocols, contribution tracking is active whenever
-# multi-agent coordination or attribution is needed.
-#
-# Trigger Conditions
-#
-# This protocol activates when the task involves:
-#   Shared File Modification: Modifying CLAUDE.md, AGENTS.md, shared configs
-#   PR Creation: "pull request", "PR", "merge request"
-#   Cross-Session Work: Multiple agents on same epic
-#   Audit Trail: Provenance, attribution, tracking
-#
-# Explicit Override: --protocol contribution flag on task creation.
-#
-# Requirements (RFC 2119)
-#
-# MUST:
-#   CONT-001: MUST follow commit message conventions
-#   CONT-002: MUST include provenance tags in code comments
-#   CONT-003: MUST pass all validation gates before merge
-#   CONT-004: MUST document decisions with rationale
-#   CONT-005: MUST flag conflicts with other sessions
-#   CONT-006: MUST write contribution record to manifest
-#   CONT-007: MUST set agent_type: "implementation" in manifest
-#
-# SHOULD:
-#   CONT-010: SHOULD include test coverage for changes
-#   CONT-011: SHOULD link to related tasks and research
-#   CONT-012: SHOULD document rejected alternatives
-#   CONT-013: SHOULD request review for significant changes
-#
-# MAY:
-#   CONT-020: MAY batch related changes into single contribution
-#   CONT-021: MAY defer documentation updates
-#   CONT-022: MAY propose follow-up improvements
-#
-# Commit Message Format:
-#   <type>(<scope>): <summary>
-#   Types: feat, fix, docs, test, refactor, chore, perf
-#
-# Validation Gates:
-#   Schema: JSON Schema validation (MUST pass)
-#   Tests: All tests pass (MUST pass)
-#   Lint: Code style compliance (SHOULD pass)
-#   Security: No secrets committed (MUST pass)
-#   Conflicts: No unresolved conflicts (MUST resolve)
-#
-# Provenance Thresholds:
-#   New code: 100%
-#   Existing code: 80%
-#   Legacy code: 50%
-#
-# Exit Codes:
-#   EXIT_PROTOCOL_CONTRIBUTION (65) - Contribution protocol violation
-#
-# Anti-Patterns:
-#   - Committing without provenance (breaks audit trail)
-#   - Skipping validation gates (quality regression)
-#   - Ignoring conflicts (creates merge debt)
-#   - Large unfocused commits (hard to review/revert)
-#   - Missing decision documentation (lost context)

package/skills/ct-master-tac/bundled/protocols/decomposition.cant

@@ -1,102 +0,0 @@
----
-kind: protocol
-version: 2.0.0
-id: DCMP
-title: "Decomposition Protocol"
-status: active
-type: conditional
-audience: "llm-agent, orchestrator"
-tags: "decomposition, hierarchy, atomicity, planning"
-skillRef: ct-epic-architect
-lastUpdated: 2026-04-08
-provenanceTask: T3155
-enforcement: strict
-stage: "decomposition"
-consult-when: "Breaking a large initiative into atomic tasks; creating epics; planning parallel wave execution; any task with 'decompose', 'break down', 'epic architect', or 'plan tasks'"
-input:
-  epic_title: "string — high-level initiative title"
-  scope: "string — description of the work to decompose"
-  constraints: "string[] — known constraints or exclusions"
-output:
-  tasks: "object[] — array of {id, title, size, dependencies, wave} atomic tasks"
-  dependency_graph: "object — DAG of task dependencies"
-  wave_plan: "object[] — array of {wave, tasks[]} parallel execution waves"
----
-
-# Decomposition Protocol
-#
-# Provenance: @task T3155, @epic T3147
-# Type: Conditional Protocol
-# Max Active: 3 protocols (including base)
-#
-# Trigger Conditions
-#
-# This protocol activates when the task involves:
-#   Epic Planning: "epic", "project", "initiative"
-#   Breakdown: "decompose", "break down", "split"
-#   Planning: "plan", "roadmap", "phases"
-#   Analysis: "scope", "estimate complexity"
-#
-# Explicit Override: --protocol decomposition flag on task creation.
-#
-# Hard Invariants (always enforced):
-#   - Parent must exist: parent task ID must be present in the task store
-#   - No circular references: moving a task under its own descendant is rejected
-#   - Dependency DAG integrity: blockedBy/blocks must remain acyclic
-#
-# Configurable Limits (profile-driven):
-#   maxSiblings: default 0 (unlimited for llm-agent-first); 7 for human-cognitive
-#   maxActiveSiblings: default 32 (operational concurrency guard)
-#   maxDepth: default 3 (epic -> task -> subtask)
-#
-# Enforcement Profiles:
-#   llm-agent-first (default): maxSiblings=0, maxActiveSiblings=32, maxDepth=3
-#   human-cognitive: maxSiblings=7, maxActiveSiblings=3, maxDepth=3
-#
-# Requirements (RFC 2119)
-#
-# MUST:
-#   DCMP-001: MUST follow MECE principle (Mutually Exclusive, Collectively Exhaustive)
-#   DCMP-002: MUST map dependencies between tasks
-#   DCMP-003: MUST respect the configured hierarchy.maxDepth policy (default: 3)
-#   DCMP-004: MUST verify atomicity for leaf tasks
-#   DCMP-005: MUST NOT include time estimates (use size: small/medium/large)
-#   DCMP-006: MUST include acceptance criteria for each task
-#   DCMP-007: Siblings MUST respect the configured hierarchy.maxSiblings policy
-#   DCMP-008: Depth MUST respect the configured hierarchy.maxDepth policy
-#   DCMP-009: MUST set agent_type: "analysis" in manifest
-#
-# SHOULD:
-#   DCMP-010: SHOULD identify parallel execution opportunities
-#   DCMP-011: SHOULD flag unclear requirements for HITL
-#   DCMP-012: SHOULD consider existing task overlap
-#   DCMP-013: SHOULD assign phase to each task
-#
-# MAY:
-#   DCMP-020: MAY propose multiple decomposition options
-#   DCMP-021: MAY identify risks and mitigations
-#   DCMP-022: MAY suggest task consolidation
-#
-# Atomicity Criteria (6-Point Test):
-#   1. Single File Scope: Affects <=3 tightly-coupled files
-#   2. Single Cognitive Concern: One bounded concern per task
-#   3. Clear Acceptance Criteria: Testable completion condition
-#   4. No Context Switching: Completable within a single agent session
-#   5. No Hidden Sub-Decisions: All choices made at decomposition
-#   6. Programmatic Validation: Result verifiable by code/test
-#
-# Size Definitions:
-#   small: 1-2 files, single concern, MUST be atomic
-#   medium: 3-7 files, related concerns, SHOULD be atomic
-#   large: 8+ files, multiple concerns, MUST decompose further
-#
-# Exit Codes:
-#   EXIT_PROTOCOL_DECOMPOSITION (63) - Decomposition protocol violation
-#
-# Anti-Patterns:
-#   - Time estimates (cannot predict duration accurately)
-#   - Non-atomic leaf tasks (create hidden work)
-#   - Missing dependencies (incorrect execution order)
-#   - Exceeding configured maxDepth (complexity explosion)
-#   - Overlapping scopes (not MECE - duplicate work)
-#   - Skipping atomicity check (quality regression)

package/skills/ct-master-tac/bundled/protocols/implementation.cant

@@ -1,77 +0,0 @@
----
-kind: protocol
-version: 1.0.0
-id: IMPL
-title: "Implementation Protocol"
-status: active
-type: base
-audience: "llm-agent"
-tags: "implementation, coding, development"
-skillRef: ct-task-executor
-lastUpdated: 2026-04-08
-enforcement: strict
-stage: "implementation"
-consult-when: "Implementing code, fixing bugs, building features, writing programs; any task with 'implement', 'build', 'create', 'fix', 'patch'; IVT loop implementation phase"
-input:
-  task_id: "string — task being implemented"
-  spec_ref: "string | null — path or ID of the governing specification"
-  acceptance_criteria: "string[] — conditions that must be true on completion"
-output:
-  files_changed: "string[] — list of modified/created files"
-  tests_added: "string[] — new test files or test names"
-  manifest_entry: "object — implementation summary for pipeline_manifest via cleo manifest append"
----
-
-# Implementation Protocol
-#
-# Provenance: @task T3155, @epic T3147
-# Type: Conditional Protocol
-# Max Active: 3 protocols (including base)
-#
-# Trigger Conditions
-#
-# This protocol activates when the task involves:
-#   Building: "implement", "build", "create", "develop"
-#   Coding: "code", "write", "program"
-#   Fixing: "fix", "bug", "patch", "repair"
-#   Enhancement: "improve", "enhance", "optimize"
-#
-# Explicit Override: --protocol implementation flag on task creation.
-#
-# Requirements (RFC 2119)
-#
-# MUST:
-#   IMPL-001: MUST include tests for new functionality
-#   IMPL-002: MUST follow project code style conventions
-#   IMPL-003: MUST include JSDoc/docstring provenance tags
-#   IMPL-004: MUST verify changes pass existing tests
-#   IMPL-005: MUST document breaking changes
-#   IMPL-006: MUST write implementation summary to manifest
-#   IMPL-007: MUST set agent_type: "implementation" in manifest
-#
-# SHOULD:
-#   IMPL-010: SHOULD add inline comments for complex logic
-#   IMPL-011: SHOULD refactor duplicated code
-#   IMPL-012: SHOULD update related documentation
-#   IMPL-013: SHOULD consider error handling edge cases
-#
-# MAY:
-#   IMPL-020: MAY propose architectural improvements
-#   IMPL-021: MAY add performance benchmarks
-#   IMPL-022: MAY suggest follow-up enhancements
-#
-# Provenance Thresholds:
-#   New code: 100% coverage required
-#   Existing code: 80% coverage required
-#   Legacy code: 50% coverage required
-#
-# Exit Codes:
-#   EXIT_PROTOCOL_IMPLEMENTATION (64) - Implementation protocol violation
-#
-# Anti-Patterns:
-#   - Code without tests (regression risk)
-#   - Missing provenance (lost attribution)
-#   - Skipping validation (quality regression)
-#   - Undocumented breaking changes (surprise failures)
-#   - No error handling (silent failures)
-#   - Hardcoded values (maintenance burden)

package/skills/ct-master-tac/bundled/protocols/provenance.cant

@@ -1,100 +0,0 @@
----
-kind: protocol
-version: 1.0.0
-id: PROV
-title: "Provenance Protocol"
-status: active
-type: cross-cutting
-audience: "llm-agent, orchestrator"
-tags: "provenance, traceability, lineage"
-skillRef: ct-provenance-keeper
-lastUpdated: 2026-04-08
-enforcement: advisory
-stage: "release"
-consult-when: "Generating in-toto attestations, SLSA provenance, or SBOMs; signing artifacts; any task with 'provenance', 'attest', 'sbom', 'cosign', 'checksum', or 'supply chain'; invoked by artifact-publish at signing step"
-input:
-  artifact_paths: "string[] — paths to built artifacts requiring provenance"
-  source_commit: "string — git SHA of the source commit"
-  build_environment: "string — description of build environment"
-output:
-  attestation_path: "string — path to in-toto attestation JSON"
-  sbom_path: "string | null — path to CycloneDX or SPDX SBOM"
-  slsa_level: "1 | 2 | 3 | 4 — achieved SLSA Build Level"
-  digests: "object — map of artifact_path to sha256 digest"
----
-
-# Provenance Protocol
-#
-# Provenance: @task T428 (gap patch)
-# Type: Conditional Protocol
-# Max Active: 3 protocols (including base)
-#
-# Trigger Conditions
-#
-# This protocol activates when the task involves:
-#   Supply Chain: "provenance", "supply chain", "chain of custody"
-#   Attestation: "attest", "attestation", "in-toto", "SLSA"
-#   SBOM: "sbom", "bill of materials", "cyclonedx", "spdx"
-#   Signing: "sign", "cosign", "sigstore", "verify signature"
-#   Checksums: "checksum", "digest", "sha256", "integrity"
-#
-# Explicit Override: --protocol provenance flag on task creation.
-#
-# Requirements (RFC 2119)
-#
-# MUST:
-#   PROV-001: MUST record provenance chain from source commit to published artifact
-#   PROV-002: MUST compute SHA-256 digest for every produced artifact
-#   PROV-003: MUST generate attestation in in-toto Statement v1 format
-#   PROV-004: MUST record SLSA Build Level achieved (L1 minimum)
-#   PROV-005: MUST store provenance record in .cleo/releases.json via record_release()
-#   PROV-006: MUST verify provenance chain integrity before publishing attestation
-#   PROV-007: MUST set agent_type: "provenance" in manifest
-#
-# SHOULD:
-#   PROV-010: SHOULD generate SBOM (CycloneDX or SPDX) for artifacts with dependencies
-#   PROV-011: SHOULD sign attestations using keyless signing (sigstore/cosign)
-#   PROV-012: SHOULD publish provenance attestation alongside artifact
-#   PROV-013: SHOULD verify all input materials have provenance
-#
-# MAY:
-#   PROV-020: MAY achieve SLSA Build Level 3 or 4
-#   PROV-021: MAY use key-based signing (GPG) as alternative to keyless
-#   PROV-022: MAY generate multiple SBOM formats (both CycloneDX and SPDX)
-#
-# Provenance Chain Model:
-#   commit -> build -> artifact -> attestation -> registry
-#
-# Chain Integrity Rules:
-#   Each link MUST reference previous link's output
-#   No link MAY be modified after creation (append-only)
-#   Missing links MUST be recorded as incomplete
-#   Chain MUST be verifiable offline
-#
-# SLSA Compliance Levels:
-#   L1: Provenance exists
-#   L2: Provenance signed + build on hosted platform
-#   L3: Non-falsifiable provenance
-#   L4: All deps have provenance + hermetic reproducible build
-#
-# Signing Methods:
-#   sigstore (default): cosign sign-blob --yes <artifact>
-#   gpg: gpg --detach-sign --armor -u <key-id> <artifact>
-#   none: Skip signing (SLSA L1 only)
-#
-# Error Codes (90-94):
-#   90: E_PROVENANCE_CONFIG_INVALID
-#   91: E_SIGNING_KEY_MISSING
-#   92: E_SIGNATURE_INVALID
-#   93: E_DIGEST_MISMATCH
-#   94: E_ATTESTATION_INVALID
-#
-# Anti-Patterns:
-#   - Skipping digest computation (breaks chain integrity)
-#   - Hardcoding signing keys in config (security risk)
-#   - Generating attestation without matching digest
-#   - Publishing artifact before signing
-#   - Modifying provenance records after creation
-#   - Skipping SBOM for artifacts with dependencies
-#   - Using SHA-1 or MD5 for digests (cryptographically broken)
-#   - Storing private keys in .cleo/ directory