contextcompass 2.15.0__py3-none-any.whl

contextcompass_cli/_package/agent_onboarding/default/new/README.md

@@ -0,0 +1,54 @@
+
+
+# Default Profile: new
+
+Purpose
+- First-time onboarding profile.
+- Minimal path to onboard a developer before selecting a steady-state profile.
+
+Rules
+- Keep this profile lightweight and orientation-focused.
+- Do not place deep language-specific execution rules here.
+- On completion, route to `general` or `engineer` by config.
+- Keep this profile user-facing: explain system purpose, classes, and config.
+
+Folder structure
+- `skills/`: first-time onboarding skills.
+- `policies/`: onboarding policy and completion rules.
+- `behavioral_guidelines/`: user-onboarding behavior flow.
+- `examples/`: reserved for first-time onboarding examples.
+
+First-time setup focus
+- Use this profile to select a steady-state role.
+- The registry table in `context_compass/SKILLS.MD` is the authoritative role
+  list. The summary below is orientation material and may lag the registry; if
+  the two disagree, the registry wins.
+- `general`: shared system/policy/ticketing baseline.
+- `engineer`: programming/testing specialization layered on `general`.
+- Fiction-authoring specializations layered on `general`:
+  - `story_designer`
+  - `story_novel_artist`
+  - `researcher`
+  - `draft_writer`
+  - `developmental_editor`
+  - `line_copy_editor`
+  - `continuity_fact_checker`
+  - `proofreader`
+- `synaptic_python_developer`: user-defined overlay layered on `engineer`.
+
+Agent-read policy
+- README reads are allowed for `new` first-time onboarding.
+- Non-new profile execution should use map/policy docs, not README files.
+
+SKILLS.MD top-level sources
+- `context_compass/SKILLS.MD`
+- `context_compass/agent_onboarding/default/new/SKILLS.MD`
+
+Primary onboarding docs
+- `agent_onboarding/default/new/skills/system_overview_for_user.md`
+- `agent_onboarding/default/new/skills/profile_model_explained.md`
+- `agent_onboarding/default/new/skills/configuration_map_guide.md`
+- `agent_onboarding/default/new/skills/onboarding_completion_and_next_step.md`
+- `PROFILE_CLASS_CREATION_GUIDE.md` (root user guide; mapped only in `new`)
+
+

contextcompass_cli/_package/agent_onboarding/default/new/SKILLS.MD

@@ -0,0 +1,35 @@
+
+
+# SKILLS.MD - new
+
+Purpose
+- Define active onboarding skills for the `new` role.
+
+Inheritance
+- `INHERITS_SKILLS_FROM: none`
+
+Activation rule
+- All paths listed below are active skills for this role.
+- YOU MUST READ ALL PATHS BELOW for all these skills to be considered active.
+
+Active skills
+- `agent_onboarding/default/new/AGENTS.MD`
+- `agent_onboarding/default/new/skills/new.md`
+- `agent_onboarding/default/general/skills/execution_contract.md`
+- `config/context_compass_config.yaml`
+- `SKILLS.MD`
+- `agent_onboarding/default/new/policies/new_onboarding_policy.md`
+- `agent_onboarding/default/new/behavioral_guidelines/user_onboarding_flow.md`
+- `agent_onboarding/default/new/skills/system_overview_for_user.md`
+- `agent_onboarding/default/new/skills/profile_model_explained.md`
+- `agent_onboarding/default/new/skills/configuration_map_guide.md`
+- `agent_onboarding/default/new/skills/first_time_profile_setup.md`
+- `agent_onboarding/default/new/skills/onboarding_completion_and_next_step.md`
+- `agent_onboarding/default/general/skills/role_local_workflows.md`
+- `agent_onboarding/default/new/WORKFLOWS.MD`
+- `PROFILE_CLASS_CREATION_GUIDE.md`
+- `agent_onboarding/default/new/README.md`
+
+
+
+

contextcompass_cli/_package/agent_onboarding/default/new/WORKFLOWS.MD

@@ -0,0 +1,15 @@
+# WORKFLOWS.MD - new
+
+Purpose
+- Define role-local workflow availability for the `new` role.
+
+Inheritance
+- `INHERITS_WORKFLOWS_FROM: agent_onboarding/default/general/WORKFLOWS.MD`
+
+Activation rule
+- This role inherits general workflow rules first.
+- Additional onboarding workflows, when created, are user-generated and
+  user-approved only.
+
+Active workflows
+- none

contextcompass_cli/_package/agent_onboarding/default/new/behavioral_guidelines/user_onboarding_flow.md

@@ -0,0 +1,52 @@
+
+
+# user_onboarding_flow
+
+Purpose
+- Provide the default interaction flow for onboarding a user in `new`.
+
+Flow
+1) Start with purpose and boundaries.
+   - Explain what context_compass is and what it is not.
+2) Explain the system at a high level.
+   - Tickets are durable planning memory.
+   - Attention board is active routing state.
+   - Artifacts support tickets when needed.
+   - `SKILLS.MD` declares the roles; config controls behaviour.
+3) Explain classes/profiles.
+   - Read the registry table in `context_compass/SKILLS.MD` and present the
+     roles from it; do not recite a list from this file.
+   - `new` is the first-time onboarding entry role and is never steady-state.
+   - Use the `extends` column to explain inheritance.
+4) Explain the authority split.
+   - `context_compass/SKILLS.MD` is the single role registry: which roles
+     exist, where each resolves, and which are selectable.
+   - `context_compass/config/context_compass_config.yaml` holds behaviour
+     settings only. No key in it selects or changes a role.
+5) Ask for role selection.
+   - Offer only roles whose `selectable after onboarding` column is `yes`.
+   - Recommend `engineer`.
+6) Confirm the selected role and next step.
+   - Explain that the choice is per session and is not stored anywhere; each
+     new session or re-onboarding selects a role again.
+   - Tell the user what happens after leaving `new`.
+
+Communication style
+- Be concise, explicit, and technical.
+- Avoid deep implementation detail unless the user asks for it.
+- Use concrete file paths when directing user actions.
+
+Output checklist
+- [ ] System purpose explained.
+- [ ] AI usage recommendation stated (Extra High reasoning).
+- [ ] Profile model and inheritance explained.
+- [ ] Registry vs config authority split explained.
+- [ ] Role selection completed (or pending explicit user choice).
+
+References
+- `agent_onboarding/default/new/policies/new_onboarding_policy.md`
+- `agent_onboarding/default/new/skills/system_overview_for_user.md`
+- `agent_onboarding/default/new/skills/profile_model_explained.md`
+- `agent_onboarding/default/new/skills/configuration_map_guide.md`
+
+

contextcompass_cli/_package/agent_onboarding/default/new/policies/new_onboarding_policy.md

@@ -0,0 +1,59 @@
+
+# new_onboarding_policy
+
+Purpose
+- Define the mandatory policy for `new` first-time onboarding.
+- Keep this profile user-facing and orientation-heavy, not implementation-heavy.
+
+Scope
+- Applies only when routing through `agent_onboarding/default/new/SKILLS.MD`.
+- Ends when the user chooses and applies a default steady-state class/profile.
+
+Policy
+- Explain the system in plain language before asking for configuration changes.
+- Describe the system goal accurately:
+  - context_compass is a policy-driven workflow system for code and fiction,
+  - it supports any programming language,
+  - it improves consistency across AI coding agents generally.
+- State current execution recommendation explicitly:
+  - use the strongest reasoning setting your runtime offers,
+  - this system trades tokens for reliability, so weak reasoning modes tend to
+    produce onboarding claims the agent cannot actually back.
+- Explain profile classes and inheritance before asking for selection.
+- Present the full set of roles by reading the registry table in `SKILLS.MD`.
+  Do not present a role list from any other document; only the registry is current.
+- Make `engineer` the recommended default class for general code development.
+- If the user needs a specialized posture, route them to the matching role:
+  - software lane: `design_engineer`, `platform_engineer`, `qa_engineer`,
+    `security_engineer`.
+  - fiction lane: `story_designer`, `story_novel_artist`, `researcher`,
+    `draft_writer`, `developmental_editor`, `line_copy_editor`,
+    `continuity_fact_checker`, `proofreader`.
+- Keep `new` profile content minimal; do not load deep engineering policy here.
+
+Configuration authority
+- `SKILLS.MD` is the source of truth for:
+  - which roles exist,
+  - the role -> `SKILLS.MD` path map,
+  - which roles are selectable after onboarding,
+  - which roles read README files.
+- `config/context_compass_config.yaml` is the source of truth for:
+  - onboarding defaults and transitions (`profiles.onboarding.*`),
+  - workflow, artifact, formatting, and read-limit behaviour.
+  - It does NOT enumerate roles.
+- `agent_onboarding/*/SKILLS.MD` headers are the source of truth for:
+  - inheritance chain,
+  - resolved parent-first read order.
+
+Completion criteria
+- User understands what the system is for.
+- User understands profiles/classes and inheritance.
+- User understands where config lives and how to change it.
+- User is offered default class selection with `engineer` recommended.
+- Next step is clear and explicit before leaving `new`.
+
+References
+- `AGENTS.MD`
+- `SKILLS.MD`
+- `agent_onboarding/default/new/skills/first_time_profile_setup.md`
+- `agent_onboarding/default/new/skills/configuration_map_guide.md`

contextcompass_cli/_package/agent_onboarding/default/new/skills/configuration_map_guide.md

@@ -0,0 +1,83 @@
+
+# configuration_map_guide
+
+Purpose
+- Explain where configuration lives, and why role routing is not controlled
+  from it.
+
+## Two files, two jobs
+
+Role identity and routing
+- `SKILLS.MD` - the **single role registry**.
+- The registry table is the only place a role is declared. It carries the role
+  name, its `SKILLS.MD` path, its parent role, whether it is user-defined,
+  whether it is selectable after onboarding, and whether it reads READMEs.
+- A role exists if and only if it has a row in that table.
+
+Behaviour settings
+- `config/context_compass_config.yaml` - **behaviour only**.
+- It does not enumerate roles and is never consulted to discover or resolve a
+  role. Do not look for a roles map, an available-profiles list, or a readme
+  policy here; none of them exist.
+
+## Config sections
+
+- `profiles.onboarding`
+  - first-time onboarding state and transition defaults.
+- `system_of_record`
+  - whether Context Compass is the only permitted place to track work.
+- `workflow`
+  - ticket microcycle and note behaviour controls.
+- `artifacts`
+  - artifact board and lifecycle controls.
+- `documentation_format`
+  - line length and evidence formatting rules.
+- `reading`
+  - per-read line limits for chunked reading.
+
+## Keys that matter for onboarding
+
+- `profiles.onboarding.first_time_enabled`
+  - Whether first-time onboarding still needs to run.
+- `profiles.onboarding.first_time_default_profile`
+  - Entry role for first-time onboarding (typically `new`).
+- `profiles.onboarding.post_onboarding_profile_mode`
+  - `choose` = ask the user which role to take once onboarding completes.
+- `profiles.onboarding.fallback_post_onboarding_profile`
+  - Safe fallback if no explicit choice is made.
+
+Which roles may be chosen after onboarding is **not** a config key. It is the
+`selectable after onboarding` column in the `SKILLS.MD` registry.
+
+## Role assignment basics
+
+1. Confirm the role has a row in the `SKILLS.MD` registry table.
+2. Confirm the `skills path` in that row points at a file that exists.
+3. Confirm the row's `extends` value matches the `INHERITS_SKILLS_FROM` header
+   inside the role's own `SKILLS.MD`.
+
+If a directory exists under `agent_onboarding/user_defined/` but has no
+registry row, it is not a role and is not selectable.
+
+## Recommended defaults after onboarding
+
+- For general code-development work: `engineer`.
+- For specialized posture, choose the closest matching role from the registry
+  table. The table's `extends` column shows what each role builds on, so a role
+  extending `engineer` carries all engineering baseline behaviour plus its own
+  delta.
+
+## Validation checks
+
+- Read `context_compass/SKILLS.MD` and confirm the registry table parses and
+  every `skills path` resolves.
+- For the selected role, read its `SKILLS.MD` and walk `INHERITS_SKILLS_FROM`
+  to the root, confirming each parent file exists.
+- Confirm `context_compass/config/context_compass_config.yaml` contains no role
+  lists. If it does, the registry has been duplicated and must be collapsed
+  back to `SKILLS.MD`.
+
+References
+- `SKILLS.MD`
+- `agent_onboarding/default/new/skills/profile_model_explained.md`
+- `PROFILE_CLASS_CREATION_GUIDE.md`

contextcompass_cli/_package/agent_onboarding/default/new/skills/first_time_profile_setup.md

@@ -0,0 +1,60 @@
+
+# first_time_profile_setup
+
+Purpose
+- Define first-time onboarding steps before steady-state profile execution.
+- Centralize setup guidance for user orientation, profile selection, and default class assignment.
+
+First-time setup sequence
+1) Start from `AGENTS.MD` bootstrap rules.
+2) Read `SKILLS.MD` - the single role registry - and the behaviour settings in
+   `config/context_compass_config.yaml`.
+3) Explain the system purpose and onboarding model using:
+   - `system_overview_for_user.md`
+   - `profile_model_explained.md`
+   - `configuration_map_guide.md`
+4) Offer role selection for post-onboarding:
+   - Read the registry table in `context_compass/SKILLS.MD`.
+   - Offer only roles whose `selectable after onboarding` column is `yes`.
+   - Use the `extends` column to explain what each role builds on.
+   - Recommend `engineer` for most code-development execution.
+   - Describe an unfamiliar role from the Purpose section of its own
+     `SKILLS.MD` rather than from a list held in this document.
+5) Confirm first-time completion and switch to selected default profile path map.
+
+Describing roles to the user
+- The registry table is the list; each role's own `SKILLS.MD` is the
+  description. Read the role's Purpose section when the user asks what a role
+  does.
+- Do not maintain a role-description list in this file. Earlier versions did,
+  and it silently went stale every time a role was added.
+- Shape of the model, which is stable even as roles change:
+  - `general` is the shared system, ticketing, policy, and workflow baseline.
+  - `engineer` layers implementation practice on top of `general`.
+  - The specialized software roles layer deeper posture on top of `engineer`.
+  - The fiction-authoring roles layer on top of `general`.
+  - User-defined roles are project or team overlays, usually on `engineer`.
+
+User-facing recommendation
+- This system is designed for code development and supports any language.
+- It enhances AI-assisted workflows by enforcing durable context.
+- Use the strongest reasoning setting your runtime offers.
+- Other reasoning modes are currently untested in this repo.
+
+Rules
+- Keep README reads scoped to first-time setup (`new` profile).
+- Do not require README reads for non-new profile onboarding paths.
+- Treat the `SKILLS.MD` registry table as authoritative for which roles exist
+  and which are selectable.
+- Treat config YAML as authoritative for onboarding and workflow behaviour only;
+  it does not enumerate roles.
+- Treat skill-map `INHERITS_SKILLS_FROM` headers as authoritative for
+  inheritance order.
+
+References
+- `context_compass/AGENTS.MD`
+- `context_compass/SKILLS.MD` (role registry - the current role list)
+- `context_compass/config/context_compass_config.yaml` (behaviour settings)
+- `agent_onboarding/default/general/SKILLS.MD`
+- `agent_onboarding/default/new/SKILLS.MD`
+- `agent_onboarding/default/new/policies/new_onboarding_policy.md`

contextcompass_cli/_package/agent_onboarding/default/new/skills/new.md

@@ -0,0 +1,54 @@
+
+# new skill - New Profile Onboarding Contract
+
+## PRIME DIRECTIVE - COMPACTION / POLICY RETENTION (NON-NEGOTIABLE)
+
+- Baseline certification is denied unless every path in every section the resolved
+  `SKILLS.MD` chain marks as **baseline** is read (parent-first). Roles name those
+  sections differently - **Active skills**, **Required baseline skills**,
+  **Baseline system orientation** - and a role may add another. Match on the
+  baseline label, not on a fixed list of headings.
+- On-demand skills are NOT part of baseline certification. They become mandatory ONLY when triggered by the active task.
+- Self-directed skills are NOT part of baseline certification either, and they have
+  no trigger. Read them at your own initiative when the work needs them.
+- After any compaction/handoff, assume chat memory is unreliable:
+  - You MUST re-onboard per `agent_onboarding/default/general/skills/compaction_requirements.md` before any action.
+  - You MUST NOT claim you "retained" this document; re-open it instead.
+
+## 1) Purpose
+
+- This profile is for first-time user onboarding only.
+- It is user-facing orientation, not steady-state execution.
+
+## 2) Role Map Summary
+
+The authoritative role list is the registry table in `context_compass/SKILLS.MD`.
+Read it rather than relying on the summary below, which is illustrative only and
+does not include user-defined roles.
+
+Commonly used default roles:
+- `new` (first-time onboarding)
+- `general` (shared baseline)
+- `engineer` (general-purpose implementation; recommended default)
+- `design_engineer` (software/system design; architecture planning and handoff)
+- `platform_engineer` (CI/CD, deploy, observability, ops)
+- `qa_engineer` (test strategy, quality gates, release signoff)
+- `security_engineer` (threat modeling, security review, hardening)
+- `story_designer` (fiction narrative architecture and chapter planning)
+- `story_novel_artist` (story/novel visual language, art briefs, and cover direction)
+- `researcher` (evidence-backed plausibility and source constraints)
+- `draft_writer` (full-manuscript drafting and rewrite execution)
+- `developmental_editor` (structural diagnosis and rewrite planning)
+- `line_copy_editor` (line-level prose polish and consistency)
+- `continuity_fact_checker` (canon/timeline/fact validation)
+- `proofreader` (final surface/format lock)
+- `user_defined/*` (team/personal overlays)
+
+## 3) Exit Rule
+
+- `new` must end with an explicit steady-state role selection and a clear next read path.
+
+References
+- `SKILLS.MD`
+- `agent_onboarding/default/new/AGENTS.MD`
+

contextcompass_cli/_package/agent_onboarding/default/new/skills/onboarding_completion_and_next_step.md

@@ -0,0 +1,54 @@
+
+# onboarding_completion_and_next_step
+
+Purpose
+- Define how to finish `new` onboarding cleanly and transition to steady-state.
+
+Completion sequence
+1) Confirm user understands:
+   - system purpose,
+   - profile class model,
+   - configuration authority.
+2) Present role options by reading the registry table in
+   `context_compass/SKILLS.MD` and offering every role whose
+   `selectable after onboarding` column is `yes`. The descriptions below are
+   a convenience gloss for the shipped default roles, not the list itself:
+   - `general` (shared baseline)
+   - `engineer` (recommended default)
+   - `design_engineer` (design/architecture/handoff)
+   - `platform_engineer` (CI/CD/deploy/observability/ops)
+   - `qa_engineer` (test strategy/quality gates/release signoff)
+   - `security_engineer` (security review/threat modeling/hardening)
+   - `story_designer` (fiction narrative architecture)
+   - `story_novel_artist` (fiction visual language and art direction)
+   - `researcher` (evidence-backed plausibility and constraints)
+   - `draft_writer` (manuscript drafting and rewrites)
+   - `developmental_editor` (structural editing and rewrite planning)
+   - `line_copy_editor` (line-level prose polish)
+   - `continuity_fact_checker` (canon/timeline/fact validation)
+   - `proofreader` (final lock for surface quality and formatting)
+3) Recommend `engineer` explicitly for general code-development work.
+4) If the user requests a specialized posture, recommend the matching role instead.
+5) Confirm selected class and align config guidance.
+6) State next action:
+   - continue onboarding via selected class path map.
+
+Recommended wording
+- "Onboarding is complete. Role options are `general`, `engineer`,
+  `design_engineer`, `platform_engineer`, `qa_engineer`, `security_engineer`,
+  `story_designer`, `story_novel_artist`, `researcher`, `draft_writer`,
+  `developmental_editor`, `line_copy_editor`, `continuity_fact_checker`,
+  or `proofreader`.
+  For general code development, `engineer` is recommended. Which do you want to
+  use for this session?"
+
+Exit criteria
+- Selected role is explicit.
+- User is informed a role is selected at each onboarding, not stored in config.
+- Next read path is clear and deterministic.
+
+References
+- `agent_onboarding/default/new/skills/configuration_map_guide.md`
+- `config/context_compass_config.yaml`
+- `SKILLS.MD`
+

contextcompass_cli/_package/agent_onboarding/default/new/skills/profile_model_explained.md

@@ -0,0 +1,120 @@
+
+# profile_model_explained
+
+Purpose
+- Explain profile classes, inheritance, and how profile routing works.
+
+Profile classes
+- `new`
+  - First-time onboarding path.
+  - Focuses on user orientation and setup.
+
+> The registry table in `context_compass/SKILLS.MD` is the authoritative
+> role list. The descriptions below are teaching material for first-time
+> onboarding and do not include user-defined roles. If the two disagree,
+> the registry wins.
+
+- `general`
+  - Shared system mechanics and workflow behavior.
+  - Baseline class for all work.
+- `engineer`
+  - General-purpose implementation specialization layered on top of `general`.
+  - Includes engineering quality, debugging discipline, and system-doc mechanics.
+- `design_engineer`
+  - Software/system design specialization layered on top of `engineer`.
+  - Includes architecture planning, decomposition, ADR hygiene, and design handoff.
+- `platform_engineer`
+  - Platform/operations specialization layered on top of `engineer`.
+  - Includes CI/CD, deployments, observability, incident workflow, and production safety.
+- `qa_engineer`
+  - Quality/test specialization layered on top of `engineer`.
+  - Includes test strategy, test design, automation practices, and release signoff mechanics.
+- `security_engineer`
+  - Security specialization layered on top of `engineer`.
+  - Includes threat modeling, secure design reviews, dependency risk, and vulnerability handling.
+- `story_designer`
+  - Fiction narrative-architecture specialization layered on top of `general`.
+  - Includes premise design, arc architecture, chapter mapping, and story-bible design.
+- `story_novel_artist`
+  - Fiction visual-language specialization layered on top of `general`.
+  - Includes style systems, scene art briefs, and cover-direction constraints.
+- `researcher`
+  - Evidence and plausibility specialization layered on top of `general`.
+  - Includes source quality triage, confidence labeling, and constraint synthesis.
+- `draft_writer`
+  - Manuscript drafting specialization layered on top of `general`.
+  - Includes draft completion, rewrite execution, and deviation logging.
+- `developmental_editor`
+  - Structural editing specialization layered on top of `general`.
+  - Includes pacing/stakes/arc diagnosis and rewrite-plan generation.
+- `line_copy_editor`
+  - Prose-quality specialization layered on top of `general`.
+  - Includes clarity polishing, style-sheet management, and mechanical cleanup.
+- `continuity_fact_checker`
+  - Canon/timeline/fact integrity specialization layered on top of `general`.
+  - Includes continuity matrices, conflict classification, and fact-risk escalation.
+- `proofreader`
+  - Final surface-quality specialization layered on top of `general`.
+  - Includes typo/punctuation/format lock and final waiver logging.
+- `user_defined/<profile_name>`
+  - Optional overlay class for personal or team preferences.
+  - Should extend defaults instead of replacing them.
+
+Inheritance model
+- `engineer` extends `general`.
+- `design_engineer` extends `engineer`.
+- `platform_engineer` extends `engineer`.
+- `qa_engineer` extends `engineer`.
+- `security_engineer` extends `engineer`.
+- `story_designer` extends `general`.
+- `story_novel_artist` extends `general`.
+- `researcher` extends `general`.
+- `draft_writer` extends `general`.
+- `developmental_editor` extends `general`.
+- `line_copy_editor` extends `general`.
+- `continuity_fact_checker` extends `general`.
+- `proofreader` extends `general`.
+- User-defined classes typically extend the most specific default role they need:
+  - e.g. extend `engineer` for language/style preferences,
+  - extend `platform_engineer` for infra-team overlays,
+  - extend `security_engineer` for org-specific security controls.
+- Read order is parent first, child last.
+- Child maps should not duplicate parent paths.
+
+Class selection model
+- First-time entry uses `new`.
+- After onboarding, the user chooses a steady-state role.
+- Recommended default for general development workflows: `engineer`.
+- Choose specialized roles when the task requires deeper domain posture.
+- Selection is per agent, per session. It is not written anywhere and not
+  shared between agents. Two agents in the same repository may hold different
+  roles at the same time, so there is no single stored "current" role.
+
+Where roles are declared
+- `SKILLS.MD` - the single role registry.
+  - One row per role: name, `SKILLS.MD` path, parent, user-defined flag,
+    selectable-after-onboarding flag, README flag.
+  - A role exists if and only if it has a row there.
+- `config/context_compass_config.yaml` holds behaviour settings only. It does
+  not enumerate roles. The only onboarding keys it carries are
+  `profiles.onboarding.*`.
+
+Where inheritance is defined
+- The registry `extends` column names the parent role.
+- The authoritative declaration lives in the role's own `SKILLS.MD` header:
+  - `` - `INHERITS_SKILLS_FROM: <skills_path|none>` ``
+- The two must agree. The header is what an agent walks; the column is what a
+  human reads.
+- Parent `SKILLS.MD` paths are loaded before child `SKILLS.MD` paths.
+
+Rules for custom classes
+- Keep shared process in `general`.
+- Keep generalized engineering mechanics in `engineer`.
+- Keep specialized domain posture in the closest matching default role.
+- Keep personal/team bias in `user_defined`.
+- Avoid overlap with inherited parent paths.
+
+References
+- `agent_onboarding/default/new/skills/configuration_map_guide.md`
+- `PROFILE_CLASS_CREATION_GUIDE.md`
+

contextcompass_cli/_package/agent_onboarding/default/new/skills/system_overview_for_user.md

@@ -0,0 +1,51 @@
+
+
+# system_overview_for_user
+
+Purpose
+- Give first-time users a concise explanation of what this system does.
+
+What context_compass is
+- A workflow and context system for structured AI work.
+- A structure for planning, execution, and handoff consistency.
+- A way to improve quality and continuity when working with AI agents.
+
+What it is for
+- Durable ticket-based planning (`epic`, `story`, `task`).
+- Deterministic active routing via `attention_board.md`.
+- Controlled artifact linkage via `artifact_board.md`.
+- Profile-based onboarding paths via config + skill maps.
+
+Language support
+- Designed for code development in any language and policy-driven fiction
+  authoring workflows.
+- The profile model allows domain specialization without changing the shared
+  core process.
+
+AI usage model
+- Primary target: AI-assisted development workflows, any runtime.
+- Compatible with other AI agents when they follow the same policy contracts.
+- Recommended execution mode in this repo:
+  - the strongest reasoning setting your runtime offers.
+  - Other reasoning modes are currently untested in this repo context.
+
+Core system anchors
+- `AGENTS.MD`
+- `config/context_compass_config.yaml`
+- `SKILLS.MD`
+- `tickets/`
+- `attention_board.md`
+- `artifact_board.md`
+
+Outcome for user
+- You should understand:
+  - why the system exists,
+  - where state lives,
+  - how onboarding and profile routing work.
+
+References
+- `agent_onboarding/default/new/skills/profile_model_explained.md`
+- `agent_onboarding/default/new/skills/configuration_map_guide.md`
+
+
+

contextcompass_cli/_package/agent_onboarding/default/new/workflows/README.md

@@ -0,0 +1,8 @@
+# new workflows
+
+Purpose
+- Store role-local workflows for the `new` role.
+
+Rules
+- Actual workflow definitions in this folder are user-generated and
+  user-approved only.