@aryaminus/controlkeel-opencode 0.2.17 → 0.2.19

package/.opencode/skills/policy-training/SKILL.md

@@ -0,0 +1,35 @@
+---
+name: policy-training
+description: "Train, inspect, promote, and archive ControlKeel router and budget-hint policy artifacts. Use this only for operator-initiated policy work."
+license: Apache-2.0
+compatibility:
+  - codex
+  - claude-standalone
+  - claude-plugin
+  - copilot-plugin
+  - github-repo
+  - open-standard
+disable-model-invocation: true
+metadata:
+  author: controlkeel
+  version: "2.0"
+  category: policy
+  ck_mcp_tools:
+    - ck_outcome_tracker
+---
+
+# Policy Training Skill
+
+Use this skill only when the task is offline policy training or artifact promotion.
+
+## Workflow
+
+1. Confirm whether you are training `router` or `budget_hint`.
+2. Use public and held-out benchmark data appropriately.
+3. Review promotion gates and never weaken deterministic controls.
+4. Summarize held-out metrics against the heuristic baseline before promotion.
+5. Consider real-world success inputs using `ck_outcome_tracker` (leaderboards, recorded session outcomes).
+
+## Additional resources
+
+- [Promotion rules](references/promotion-rules.md)

package/.opencode/skills/policy-training/agents/openai.yaml

@@ -0,0 +1,12 @@
+interface:
+  display_name: "Policy Training"
+  short_description: "Operate CK router and budget-hint artifact training safely."
+  brand_color: "#7c2d12"
+policy:
+  allow_implicit_invocation: false
+dependencies:
+  tools:
+    - type: "mcp"
+      value: "controlkeel"
+      description: "ControlKeel MCP server"
+

package/.opencode/skills/policy-training/references/promotion-rules.md

@@ -0,0 +1,6 @@
+# Promotion Rules
+
+- Router artifacts must improve held-out utility without degrading catch rate or expected-rule hit rate.
+- Budget-hint artifacts may raise `allow` to `warn`, but never weaken deterministic `warn` or `block`.
+- Training is offline and operator initiated; runtime inference remains pure Elixir.
+

package/.opencode/skills/proof-memory/SKILL.md

@@ -0,0 +1,39 @@
+---
+name: proof-memory
+description: "Use ControlKeel proof bundles, typed memory, workspace snapshots, transcript summaries, and resume packets before closing or resuming work. Activate this when you need durable evidence or historical context."
+license: Apache-2.0
+compatibility:
+  - codex
+  - claude-standalone
+  - claude-plugin
+  - copilot-plugin
+  - github-repo
+  - open-standard
+metadata:
+  author: controlkeel
+  version: "2.0"
+  category: proof
+  ck_mcp_tools:
+    - ck_context
+    - ck_memory_search
+    - ck_memory_record
+    - ck_memory_archive
+    - ck_regression_result
+---
+
+# Proof and Memory Skill
+
+Use this skill when you need the durable system-of-record view instead of only the live session state.
+
+## Workflow
+
+1. Call `ck_context`.
+2. Review `proof_summary`, `memory_hits`, `workspace_context`, `context_reacquisition`, `instruction_hierarchy`, `recent_events`, and `resume_packet`.
+3. Use the proof bundle to understand deploy readiness, regression evidence, open findings, and rollback expectations.
+4. Use `ck_memory_search` when prior decisions, checkpoints, or findings need explicit retrieval instead of relying only on passive memory hits.
+5. Use `ck_memory_record` to preserve new decisions or operator intent that future agents should recover explicitly.
+6. Use `ck_memory_archive` to retire stale or superseded memories so retrieval quality does not decay.
+
+## Additional resources
+
+- [Proof workflow](references/proof-workflow.md)

package/.opencode/skills/proof-memory/agents/openai.yaml

@@ -0,0 +1,11 @@
+interface:
+  display_name: "Proof and Memory"
+  short_description: "Use CK proof bundles, typed memory, workspace context, and resume packets."
+  brand_color: "#0f172a"
+policy:
+  allow_implicit_invocation: true
+dependencies:
+  tools:
+    - type: "mcp"
+      value: "controlkeel"
+      description: "ControlKeel MCP server"

package/.opencode/skills/proof-memory/references/proof-workflow.md

@@ -0,0 +1,6 @@
+# Proof Workflow
+
+- Before completion, verify that the latest proof bundle reflects the current task state.
+- When resuming paused work, use the resume packet plus memory hits, explicit memory search, workspace context, context reacquisition signals, and recent transcript events before making changes.
+- If external QA systems report browser or regression failures, record them through `ck_regression_result` before treating the proof as deploy-ready.
+- Treat proof bundles as immutable evidence snapshots, not editable state.

package/.opencode/skills/security-review/SKILL.md

@@ -0,0 +1,34 @@
+---
+name: security-review
+description: "Run a structured security review before marking a task done. Use this for code, config, architecture, or release reviews that need OWASP, baseline pack, and domain-pack coverage."
+license: Apache-2.0
+compatibility:
+  - codex
+  - claude-standalone
+  - claude-plugin
+  - copilot-plugin
+  - github-repo
+  - open-standard
+metadata:
+  author: controlkeel
+  version: "2.0"
+  category: security
+  ck_mcp_tools:
+    - ck_validate
+    - ck_context
+    - ck_finding
+    - ck_regression_result
+---
+
+# Security Review Skill
+
+Use this skill before closing a task, approving a proof bundle, or reviewing a risky diff.
+
+## Review flow
+
+1. Call `ck_context` to load the domain pack, risk tier, open findings, instruction hierarchy, and design-drift signals.
+2. Run `ck_validate` on the relevant code or config slices, including trust-boundary metadata when the proposed action was influenced by web, tool, skill, or mixed-provenance content.
+3. Walk the review checklist in [references/review-checklist.md](references/review-checklist.md).
+4. Persist any missed issue with `ck_finding`.
+5. If external security or regression systems produce exploit or browser evidence, record that through `ck_regression_result` when it affects release readiness.
+6. Summarize blockers, warnings, and follow-up proof requirements.

package/.opencode/skills/security-review/agents/openai.yaml

@@ -0,0 +1,12 @@
+interface:
+  display_name: "Security Review"
+  short_description: "Structured CK security review for code, config, and release gating."
+  brand_color: "#b91c1c"
+policy:
+  allow_implicit_invocation: true
+dependencies:
+  tools:
+    - type: "mcp"
+      value: "controlkeel"
+      description: "ControlKeel MCP server"
+

package/.opencode/skills/security-review/references/review-checklist.md

@@ -0,0 +1,69 @@
+# Security Review Checklist
+
+Use this checklist before task completion, proof approval, or release review.
+Each item lists a concrete verification step. Run `ck_validate` for automated pattern
+detection, then manually verify items that require architectural or process judgment.
+
+## Baseline
+
+- **Secrets**: Verify no hardcoded API keys, tokens, passwords, private keys, or connection strings exist in source code, config files, environment files, or test fixtures. Run `ck_validate` and confirm zero `security.hardcoded_secret` findings. Check that CI/CD secret management uses vault references, not plaintext values.
+- **Injection (SQL)**: Verify all database queries use parameterized statements, Ecto-style bindings, or query builders. Search for string concatenation in SQL query construction. Run `ck_validate` and confirm zero `security.sql_injection` findings.
+- **Injection (shell)**: Verify no shell commands are constructed by interpolating user input directly into command strings. Use argument arrays or validated allowlists instead. Run `ck_validate` and confirm zero `security.shell_injection` findings.
+- **Injection (HTML/XSS)**: Verify no `innerHTML` assignments from untrusted input. Confirm Phoenix HEEx templates escape by default and any `raw` or `phx-no-curly-interpolation` usage is audited. Run `ck_validate` and confirm zero `security.xss` findings.
+
+## Authentication and Authorization
+
+- **Authentication on protected routes**: Verify every route that accesses user-scoped data requires authentication. Confirm middleware or plugs check session or token validity before controller action. Check that API endpoints validate bearer tokens against active, non-expired, non-revoked credentials.
+- **Authorization checks**: Verify each protected endpoint checks the caller's role or scope before performing the action. Confirm ownership checks (user can only access their own resources) are enforced at the data layer, not just the UI layer. Check that admin-only endpoints are not reachable by non-admin tokens even with direct API calls.
+- **Session management**: Verify session tokens expire within a reasonable timeframe. Confirm session invalidation on password change and logout. Check that concurrent session limits are enforced if applicable.
+- **Service account controls**: Verify service accounts have scoped permissions (not blanket admin). Confirm service account tokens have short expiry and support revocation. Check that service account activity is logged.
+
+## Input Validation and Output Encoding
+
+- **Input validation**: Verify all user-supplied input is validated for type, length, format, and range before processing. Confirm file uploads are validated for type and size, and scanned for malware if applicable. Check that JSON/XML parsers are configured to reject oversized or malformed payloads.
+- **Output encoding**: Verify all dynamic content rendered in HTML is contextually encoded (HTML body, attribute, JavaScript, URL). Confirm API responses use structured JSON, not string interpolation of user data. Check that error messages do not leak internal state, stack traces, or database details.
+- **File handling**: Verify uploaded files are stored outside the web root. Confirm file paths are validated to prevent directory traversal. Check that file content types are verified by inspection, not just the client-supplied MIME type.
+
+## API Security
+
+- **CORS policy**: Verify CORS is not configured as `*` for authenticated endpoints. Confirm allowed origins are explicitly enumerated. Check that credential-bearing requests are not sent to overly broad origins.
+- **Rate limiting**: Verify rate limiting is active on authentication, registration, and password-reset endpoints. Confirm rate limits are scoped per identity (not just per IP) where applicable. Check that rate-limit headers are not informative to attackers.
+- **Request size limits**: Verify maximum request body size is configured and enforced. Confirm file upload size limits are applied at the reverse proxy and application level. Check that JSON payload depth is bounded.
+- **TLS enforcement**: Verify all production endpoints redirect HTTP to HTTPS. Confirm HSTS headers are set. Check that TLS configuration follows current best practices (TLS 1.2+, strong cipher suites).
+
+## Data Protection
+
+- **Encryption at rest**: Verify sensitive data fields (PII, financial, health, credentials) are encrypted in the database. Confirm encryption keys are managed through a vault or KMS, not stored alongside the data. Check that key rotation is supported and documented.
+- **Encryption in transit**: Verify all internal service-to-service communication uses TLS or equivalent. Confirm database connections use SSL. Check that external API calls use HTTPS and validate certificates.
+- **Logging hygiene**: Verify no sensitive data (passwords, tokens, SSNs, card numbers, health data) appears in application logs. Confirm log levels in production do not emit debug data that includes request bodies with PII. Check that error tracking (Sentry, etc.) is configured to scrub sensitive fields.
+- **Data masking in responses**: Verify API responses mask or omit sensitive fields for non-privileged scopes. Confirm pagination responses do not leak record counts that could enable enumeration attacks. Check that error responses do not include internal identifiers or stack traces.
+
+## Dependency and Supply Chain
+
+- **Dependency audit**: Verify all direct and transitive dependencies are at non-vulnerable versions. Run `ck_validate` or equivalent dependency scanner. Confirm a process exists for timely patching of critical CVEs. Check that dependency pinning is used in production builds.
+- **Lock file integrity**: Verify lock files are committed and not modified without review. Confirm CI verifies lock file consistency. Check that dependency resolution does not silently upgrade versions.
+- **Container image hygiene**: Verify base images are minimal and up-to-date. Confirm images do not run as root. Check that no unnecessary packages or shells are included in production images.
+
+## Infrastructure and Deployment
+
+- **Environment separation**: Verify production credentials are not used in development or staging environments. Confirm environment-specific configuration is loaded from vault or environment variables, not hardcoded. Check that test fixtures use synthetic data, not production data.
+- **Secret rotation**: Verify database credentials, API keys, and signing keys are rotated on a defined schedule. Confirm rotation does not cause downtime. Check that old credentials are invalidated promptly after rotation.
+- **Backup and recovery**: Verify backups are encrypted and access-controlled. Confirm backup restoration has been tested within the retention period. Check that backup data includes integrity checksums.
+- **Infrastructure as code**: Verify all infrastructure changes go through version-controlled configuration (not manual console changes). Confirm infrastructure configuration is reviewed before application. Check that drift detection is active.
+
+## Domain overlays
+
+Apply the relevant domain-specific checks from [domain-review-matrix.md](../domain-audit/references/domain-review-matrix.md) in addition to the above.
+
+- **Healthcare**: PHI access, encryption, minimum necessary, audit logging, breach process.
+- **Finance**: card data isolation, tokenization, immutable financial records, segregation of duties.
+- **Education**: student-record disclosure and age/consent requirements, FERPA compliance.
+- **HR**: discriminatory criteria and employee/candidate PII exposure, compensation visibility.
+- **Legal**: privilege handling, retention, and document protection, conflict checking.
+- **Marketing/Sales/Real Estate**: consent, CRM PII, fair-housing constraints, contract handling.
+- **Government**: records retention, benefits fairness, approval chain integrity.
+- **Insurance**: claims fairness, medical-adjacent privacy, denial review.
+- **E-commerce**: card scope, refund controls, fraud detection.
+- **Logistics**: custody integrity, dispatch safety, carrier data protection.
+- **Manufacturing**: QA holds, traceability, safety interlocks.
+- **Nonprofit**: donor privacy, grant restrictions, beneficiary data protection.

package/.opencode/skills/ship-readiness/SKILL.md

@@ -0,0 +1,35 @@
+---
+name: ship-readiness
+description: "Check install-to-first-finding metrics, funnel stage, findings state, proofs, and approvals before calling a session ready to ship."
+license: Apache-2.0
+compatibility:
+  - codex
+  - claude-standalone
+  - claude-plugin
+  - copilot-plugin
+  - github-repo
+  - open-standard
+metadata:
+  author: controlkeel
+  version: "2.0"
+  category: release
+  ck_mcp_tools:
+    - ck_context
+    - ck_deployment_advisor
+---
+
+# Ship Readiness Skill
+
+Use this skill when the operator asks whether a mission or session is ready for release.
+
+## Workflow
+
+1. Check session metrics and current funnel stage.
+2. Verify there are no unresolved blockers.
+3. Confirm proof state and rollback guidance.
+4. Summarize approvals, rejections, and any remaining human work.
+5. Provide automatic deployment resources via `ck_deployment_advisor` (Dockerize, CI pipes) for the relevant stack (Phoenix, etc.).
+
+## Additional resources
+
+- [Release checklist](references/release-checklist.md)

package/.opencode/skills/ship-readiness/agents/openai.yaml

@@ -0,0 +1,12 @@
+interface:
+  display_name: "Ship Readiness"
+  short_description: "Release gating and session-readiness checks for CK."
+  brand_color: "#4f46e5"
+policy:
+  allow_implicit_invocation: true
+dependencies:
+  tools:
+    - type: "mcp"
+      value: "controlkeel"
+      description: "ControlKeel MCP server"
+

package/.opencode/skills/ship-readiness/references/release-checklist.md

@@ -0,0 +1,8 @@
+# Release Checklist
+
+- No blocked findings remain unresolved.
+- Required approvals are completed.
+- Latest proof bundle is deploy-ready or clearly states why not.
+- Budget state is healthy enough for the remaining deployment workflow.
+- Ship metrics do not show an incomplete funnel stage for the active session.
+

package/package.json

@@ -35,5 +35,5 @@
     "url": "git+https://github.com/aryaminus/controlkeel.git"
   },
   "type": "module",
-  "version": "0.2.17"
+  "version": "0.2.19"
 }