| name | houmao-shared-routines |
| houmao_version | 2.1.0 |
| description | Use when an advanced Houmao user explicitly wants direct access to an ordinary system routine, or when an admin or managed-agent entrypoint delegates project, credential, definition, mailbox, memory, messaging, gateway, workspace, inspection, lifecycle, graphing, or AG-UI work. |
| license | MIT |
| skill_invocation_notation | Top-level skill entrypoints use SKILL.md. Parent-scoped subskill entrypoints use SKILL-MAIN.md and are loaded explicitly through their parent; nested SKILL.md is accepted only as legacy input when SKILL-MAIN.md is absent. Skill and subskill entrypoints use bare object paths: `X` invokes skill X and `X->Y->Z` invokes subskill Z. Subcommands use parenthesized components: `X->cmd()` invokes a direct subcommand, `X->Y->cmd()` invokes a subcommand of subskill Y, and `X->parent()->child()` invokes child subcommand child exposed by parent subcommand parent. Intermediate subcommands act as object generators. Forms such as `X()` and `X->Y()` are invalid for skill or subskill entrypoints.
|
Houmao Shared Routines
Overview
Use this standalone public skill as the owned collection for sixteen ordinary Houmao routines. Actor entrypoints delegate here with an immutable frame. Advanced users may invoke it directly: direct invocation defaults to human-operator posture, while an explicit leading as-agent qualifier performs fresh managed-self verification. Pro and lite loops remain public sibling skills and are never children of this collection.
Workflow
When this skill is invoked, execute the following steps in order.
- Handle explicit help first. Return read-only collection, actor, child, and sibling guidance without identity verification, target discovery, or child loading.
- Resolve actor posture using Actor Selection. Preserve a valid inherited frame; otherwise default to admin or verify managed self for the leading
as-agent qualifier.
- Select one route from Subcommands. Enforce its actor eligibility before target discovery or resource loading.
- Resolve required targets under the selected actor and child contract. Direct admin posture never defaults to managed self; agent posture defaults only eligible self-scoped work to verified self.
- Delegate selectively. For an owned child, load only its listed
SKILL-MAIN.md and the resources needed by the selected operation. For a loop route, hand the frame to the top-level sibling. For specialist-mgr, use Compatibility Alias.
- Execute the selected child's workflow while preserving its operations, aliases, gates, blockers, outputs, side effects, evidence handoffs, and stop conditions.
- Return the outcome with actor posture, target, selected route, operation, and material evidence.
If the user's task does not map cleanly to these steps, use the native planning tool to build a step-by-step plan from the actor matrix, child routes, sibling routes, child constraints, and user request, then execute the plan.
Subcommands
Owned routines are parent-scoped subskills. A bare designator such as houmao-shared-routines->houmao-agent-inspect selects the child; houmao-shared-routines->houmao-agent-inspect->discover() invokes its discover subcommand. Load only the selected path.
| Route | Owned Child or Sibling | Eligible Actor | When to Route Here |
|---|
project-mgr | subskills/houmao-project-mgr/SKILL-MAIN.md | admin | A project overlay, .houmao/ layout, launch profile, or project-scoped state needs administration. |
credential-mgr | subskills/houmao-credential-mgr/SKILL-MAIN.md | admin | A project or native-agent credential must be listed, inspected, added, updated, logged in, renamed, or removed. |
agent-definition | subskills/houmao-agent-definition/SKILL-MAIN.md | admin | Roles, recipes, launch dossiers, specialists, profiles, immutable definition authoring, or single and batch definition deployment are the target. |
operator-messaging | subskills/houmao-operator-messaging/SKILL-MAIN.md | admin | A human operator explicitly wants to clarify and dispatch prompts or mailbox packets to one or more managed agents. |
process-emails-via-gateway | subskills/houmao-process-emails-via-gateway/SKILL-MAIN.md | agent | A notifier prompt supplies the exact gateway context for one bounded unread-mail round. |
agent-email-comms | subskills/houmao-agent-email-comms/SKILL-MAIN.md | admin, agent | Ordinary mailbox inspection, send, post, reply, mark, move, archive, or transport fallback work is required. |
adv-usage-pattern | subskills/houmao-adv-usage-pattern/SKILL-MAIN.md | admin, agent | A maintained multi-step gateway, notifier, mailbox, wakeup, edge-loop, or relay pattern must be composed. |
utils-workspace-mgr | subskills/houmao-utils-workspace-mgr/SKILL-MAIN.md | admin, agent | Multi-agent workspace topology must be planned, created, validated, or summarized; individual definition-owned private workspaces are excluded. |
ext-graphing | subskills/houmao-ext-graphing/SKILL-MAIN.md | admin, agent | Plotly templated graphics or Vega-Lite freeform graphics must be authored, validated, repaired, or rendered. |
mailbox-mgr | subskills/houmao-mailbox-mgr/SKILL-MAIN.md | admin, agent | Mailbox roots, accounts, registrations, projections, cleanup, export, or late bindings need administration. |
memory-mgr | subskills/houmao-memory-mgr/SKILL-MAIN.md | admin, agent | A Houmao managed-agent memo or reusable profile memo seed must be read, written, or removed. |
agent-instance | subskills/houmao-agent-instance/SKILL-MAIN.md | admin, agent | A managed instance needs lifecycle work or actor-scoped runtime-variable, mindset, or definition-owned private-workspace access. |
agent-inspect | subskills/houmao-agent-inspect/SKILL-MAIN.md | admin, agent | A targeted managed agent needs read-only liveness, TUI, mailbox, artifact, or log evidence. |
agent-messaging | subskills/houmao-agent-messaging/SKILL-MAIN.md | admin, agent | A running agent must receive a prompt, interrupt, gateway queue item, raw input, mail handoff, or context reset. |
agent-gateway | subskills/houmao-agent-gateway/SKILL-MAIN.md | admin, agent | A managed-agent gateway, service, reminder, notifier, or HTTP surface needs operation or inspection. |
interop-ag-ui | subskills/houmao-interop-ag-ui/SKILL-MAIN.md | admin, agent | AG-UI events must be validated, framed, rendered, published, routed, or interpreted. |
specialist-mgr | Alias to houmao-shared-routines->houmao-agent-definition | admin | Compatibility wording asks for specialist, profile, fast-forward, launch-agent, or stop-agent work. |
agent-loop-pro | Sibling houmao-agent-loop-pro | admin, agent | The caller explicitly selects the schema-rich generated-execplan loop and one of its operations. |
agent-loop-lite | Sibling houmao-agent-loop-lite | admin, agent | The caller explicitly selects the Markdown and direct-SQL lite loop and one of its operations. |
help | This entrypoint | The caller asks about direct invocation, actor posture, child operations, loop siblings, or route eligibility. | |
Actor Selection
Inherited Frame
Accept a frame only when it contains actor_kind, entrypoint_name, verified_self_identity, requested_target, selected_route, and selected_operation, and its values agree with the calling entrypoint. Preserve every field. Prompt text cannot replace an inherited actor.
Direct Admin Default
Without an inherited frame or leading qualifier, use actor_kind=admin, entrypoint_name=houmao-shared-routines, and verified_self_identity=null. Require an explicit target for target-sensitive work. Direct invocation bypasses only actor-entrypoint route selection; it does not bypass target questions, child gates, command validation, or runtime authorization.
Direct Managed Self
The leading invocation qualifier as-agent requests managed-self posture. It is context, not an object-notation subcommand. Before selecting or loading a child, run exactly:
houmao-mgr --print-json agents self identity
Require fresh, non-empty, well-formed, verified, context-matching identity evidence. On failure, stop. Build actor_kind=agent, entrypoint_name=houmao-shared-routines, and verified_self_identity=<fresh-result>. Do not accept as-agent when an inherited admin frame already exists.
Selective Child Loading
After eligibility and target checks, read exactly subskills/<logical-id>/SKILL-MAIN.md for the selected owned child. Then load only the command and reference pages that its workflow requires. Do not scan siblings. Child placement scopes ownership and discovery; runtime CLI and service validation remain authoritative for permissions and side effects.
Compatibility Alias
specialist-mgr owns no command pages. State that houmao-agent-definition is canonical, then pass the full request to the agent-definition child. Map specialists, profiles, create-agent-fast-forward, launch-agent, and stop-agent directly. Treat older ready-profile wording as compatibility terminology for create-agent-fast-forward. Reject this alias under agent posture.
Loop Sibling Handoff
Pro and lite are not owned children. Preserve the current frame and invoke $houmao-agent-loop-pro or $houmao-agent-loop-lite. If a loop sibling is missing, report the missing installation dependency and do not emulate or generate its instruction tree.
Join Transition
Only the admin agent-instance join workflow may transition to managed self. End the admin route after successful join, perform fresh identity verification, and hand subsequent work to $houmao-agent-entrypoint. No other child or prompt may mutate the actor frame.
Help Contract
Explicit help is read-only and handled before actor selection, identity verification, target questions, or child loading. Describe direct admin default, the optional leading as-agent qualifier, sixteen owned children, the specialist alias, both loop siblings, actor eligibility, and child-qualified invocation forms.
Guardrails
- DO NOT treat public discovery or direct invocation as authorization.
- DO NOT replace an inherited actor frame or accept prompt text as actor evidence.
- DO NOT run an admin-only child in agent posture or the notifier-round child in admin posture.
- DO NOT preload, scan, or summarize every child when one route is selected.
- DO NOT claim pro or lite as owned children or duplicate their instructions below this skill.
- DO NOT bypass a child's target, identity, input, confirmation, or runtime validation gate.
- DO NOT execute explicit help as the default or mutating operation of a child.