@xevy/skills 0.1.37 → 0.1.38

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/README.md CHANGED
@@ -104,16 +104,10 @@ with `codex mcp add`.
104
104
 
105
105
  | Skill | Purpose |
106
106
  |-------|---------|
107
- | `bot` | Use a connected Bot provider as a complete agent platform |
108
- | `bot-agent` | Delegate to and collect a Bot Agent as a private subagent |
109
- | `bot-work` | Control durable Bot work, receipts, events, handoff, and artifacts |
110
- | `bot-job` | List, create, configure, run, pause, and resume Bot Jobs |
111
- | `bot-browser` | Use the selected Bot Agent Browser |
112
- | `bot-computer` | Use Bot Computer browser, files, processes, and credentials |
113
- | `bot-connector` | Run assigned Bot Connector tools |
114
- | `bot-skills` | Manage and run Heny product Skills |
115
- | `bot-channel` | Deliver through an assigned personal Bot Channel |
116
- | `bot-memory` | Read and manage confirmed Bot Agent memory |
107
+ | `bot` | Use a connected Bot through its Agent's primary conversation |
108
+ | `bot-agent` | Message a Bot Agent and read its background Activities |
109
+ | `bot-job` | Read Bot Jobs and route changes through the primary Agent conversation |
110
+ | `bot-memory` | Read confirmed Bot Agent memory and route changes through the Agent |
117
111
  | `create-artifact` | Build Drive artifacts; HTML by default, with other formats on explicit request |
118
112
  | `create-document` | Create a project document with private/shared/public visibility |
119
113
  | `create-note` | Create a private personal note, optionally published on request |
package/bin/cli.mjs CHANGED
@@ -31,6 +31,12 @@ const DEFAULT_BASE = "https://xevy.ai";
31
31
  const DEFAULT_AGENTS = ["cursor", "claude-code", "codex"];
32
32
  const RETIRED_SKILLS = [
33
33
  "analyze-here",
34
+ "bot-browser",
35
+ "bot-channel",
36
+ "bot-computer",
37
+ "bot-connector",
38
+ "bot-skills",
39
+ "bot-work",
34
40
  "plan-here",
35
41
  "execute-here",
36
42
  "fix-here",
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@xevy/skills",
3
- "version": "0.1.37",
3
+ "version": "0.1.38",
4
4
  "description": "Install Xevy agent skills and secure per-agent MCP transports for Cursor, Claude Code, Codex, and other coding agents.",
5
5
  "type": "module",
6
6
  "keywords": [
@@ -1,30 +1,26 @@
1
1
  ---
2
2
  name: bot
3
- description: Use a connected personal Bot provider as a complete external agent platform. Use whenever the user asks Xevy to work through Heny, use a Bot Agent, or combine Bot Jobs, Skills, Connectors, Computers, Browser, Channels, memory, artifacts, or approvals.
3
+ description: Use a connected personal Bot through its Agent's primary conversation. Use whenever the user asks Xevy to work through Heny, use a Bot Agent, start long-running background work, or read Bot Activities, Jobs, or confirmed memory.
4
4
  argument-hint: "[objective]"
5
- allowed-tools: "mcp__xevy__bot_list mcp__xevy__bot_agents mcp__xevy__bot_capabilities mcp__xevy__bot_delegate mcp__xevy__bot_collect mcp__xevy__bot_work_start mcp__xevy__bot_work_get mcp__xevy__bot_work_heartbeat mcp__xevy__bot_remote mcp__xevy__bot_tools mcp__xevy__bot_action mcp__xevy__bot_work_finish mcp__xevy__bot_work_cancel"
5
+ allowed-tools: "mcp__xevy__bot_list mcp__xevy__bot_agents mcp__xevy__bot_capabilities mcp__xevy__bot_message mcp__xevy__bot_activity mcp__xevy__bot_job mcp__xevy__bot_memory"
6
6
  ---
7
7
 
8
8
  # Use a Bot from Xevy
9
9
 
10
- Treat the connected Bot as an external agent platform. Heny is the first provider; select another provider when the user names one and it advertises the required capability.
10
+ Treat the connected Bot Agent's primary conversation as the only execution entrance. Heny is the first provider; select another provider when the user names one and it advertises the required conversational capability.
11
11
 
12
12
  ## Select the provider and Agent
13
13
 
14
14
  1. Call `bot_list` and keep providers whose connection is active and whose capability set covers the request.
15
15
  2. Call `bot_agents`. When several workspaces or Agents match, use an explicit user name or clear conversation context. Ask the user when the remaining choice changes the outcome.
16
- 3. Call `bot_capabilities` before starting work. Explain a missing assignment using the provider URL returned by its connection metadata.
16
+ 3. Call `bot_capabilities` before sending the request. Explain a missing assignment using the provider URL returned by its connection metadata.
17
17
 
18
- ## Choose control
18
+ ## Send the request
19
19
 
20
- - Use `bot_delegate` when the user wants the Bot to act autonomously, wants Heny as a subagent, or asks for an outcome that the Bot Agent can complete with its own model and capabilities.
21
- - Use external `bot_work_start` when the current coding agent should direct each Bot tool call, inspect receipts, or combine Bot actions with local work.
22
- - Use `bot_collect` or `bot_work_get` to retrieve status and results. Preserve the returned private link.
20
+ Call `bot_message` with the complete request and a stable idempotency key. The selected Agent decides whether to answer in the turn or start one or more background Activities. Return the Agent's response and Activity references to the user.
23
21
 
24
- ## Governed work
22
+ Use `bot_activity` to list or read Activity status and structured results. When an Activity reaches a terminal state, evaluate its result against the user's original request, explain useful findings and limitations naturally, and identify the next action.
25
23
 
26
- For external work, renew its lease with `bot_work_heartbeat`, discover current schemas with `bot_tools`, run exact names with `bot_action`, and reuse the same idempotency key only for the same action and arguments. Inspect `work_receipt` through `bot_remote` before retrying an uncertain effect.
24
+ `bot_job` and `bot_memory` are read-only. Ask the Agent through `bot_message` when a Job or confirmed memory needs to change. Browser, Computer, Connector, Channel, Skill, and other capability use remains inside the Agent's governed conversation.
27
25
 
28
- Heny enforces audiences, Agent autonomy, Decisions, Connector policy, and Computer-local permissions. Surface waiting input and Decisions with their returned link. Let the person resolve Decisions in Heny.
29
-
30
- Finish external work with `bot_work_finish`. Keep `publish` false unless the person explicitly requests publication to the Agent conversation. Cancel abandoned active work.
26
+ Heny applies the Agent-user intent permission across the person's primary conversation, Jobs, Job conversations, Activities, schedules, channels, and Xevy requests. Surface `Once`, `Always`, and `Decline` decisions and waiting input without inventing authority.
@@ -1,16 +1,16 @@
1
1
  ---
2
2
  name: bot-agent
3
- description: Discover and use a connected Bot Agent, especially when the user asks Heny to act as a subagent, delegate a task, run an Agent, or collect delegated results.
3
+ description: Discover and message a connected Bot Agent, especially when the user asks Heny to handle a request or continue longer work in a background Activity.
4
4
  argument-hint: "[agent name] [objective]"
5
- allowed-tools: "mcp__xevy__bot_list mcp__xevy__bot_agents mcp__xevy__bot_capabilities mcp__xevy__bot_delegate mcp__xevy__bot_collect mcp__xevy__bot_remote"
5
+ allowed-tools: "mcp__xevy__bot_list mcp__xevy__bot_agents mcp__xevy__bot_capabilities mcp__xevy__bot_message mcp__xevy__bot_activity"
6
6
  ---
7
7
 
8
- # Bot Agent and subagent
8
+ # Bot Agent conversation
9
9
 
10
10
  Discover active providers with `bot_list`, resolve the workspace and Agent with `bot_agents`, and confirm assignments with `bot_capabilities`. Select from the user’s explicit Agent name or the request context; ask when multiple plausible Agents would produce different results.
11
11
 
12
- Call `bot_delegate` with the complete bounded objective and a new stable idempotency key. Delegated work is private to the connected person and runs through the Bot provider’s native Agent runtime.
12
+ Call `bot_message` with the complete request and a stable idempotency key. The request enters the Agent's primary conversation. The Agent may start background Activities for longer independent work and returns their references in its response.
13
13
 
14
- Collect with `bot_collect`. Report completed results, current progress, waiting input, or a pending Decision. Give the returned private URL for any state requiring the person. Continue collection after the person confirms they resolved the wait.
14
+ Read progress and results with `bot_activity`. Report completed results, current progress, waiting input, or a pending intent Decision. After completion, evaluate the result against the user's request and explain the useful outcome, limitations, and next action.
15
15
 
16
- A native Heny Agent can use its assigned `agents.delegate` and `agents.collect` tools for nested specialist work. Heny applies its delegation depth, child-count, cycle, ownership, and audience rules.
16
+ Activities extend the primary conversation and cannot create another Activity. Jobs and external callers cannot start them.
@@ -1,20 +1,14 @@
1
1
  ---
2
2
  name: bot-job
3
- description: List, create, configure, run, pause, or resume durable Jobs through a connected Bot provider. Use whenever the user mentions a Heny Job, Bot Job, schedule, recurring Agent responsibility, or Job memory.
3
+ description: Read durable Jobs through a connected Bot provider and route requested Job changes through the owning Agent's primary conversation.
4
4
  argument-hint: "[job request]"
5
- allowed-tools: "mcp__xevy__bot_list mcp__xevy__bot_agents mcp__xevy__bot_work_start mcp__xevy__bot_job mcp__xevy__bot_work_get mcp__xevy__bot_work_finish"
5
+ allowed-tools: "mcp__xevy__bot_list mcp__xevy__bot_agents mcp__xevy__bot_job mcp__xevy__bot_message"
6
6
  ---
7
7
 
8
8
  # Bot Jobs
9
9
 
10
- Resolve the provider, workspace, and Agent before changing a Job. Use `bot_job` with:
10
+ Resolve the provider, workspace, and Agent. Use `bot_job` to list authorized Jobs and their current processes, schedules, and status.
11
11
 
12
- - `list` to read authorized Jobs and current processes, schedules, status, and memory version.
13
- - `create` with a complete configuration: name, objective, process, one or more schedules with IANA time zones, and optional memory or capability requirements.
14
- - `configure` to update the complete intended fields; include `expectedMemoryVersion` for memory changes.
15
- - `run` to queue one ready Job run.
16
- - `pause` or `resume` to change scheduled execution.
12
+ Send create, configure, run, pause, or resume requests through `bot_message` to the owning Agent's primary conversation. Include the Job ID or name, the exact requested change, and any activation instruction. Heny applies the user's global Agent-intent permission and preserves the internal Job conversation.
17
13
 
18
- Create and modify Jobs through active external Bot work, passing its workspace and work IDs in the operation arguments. New Jobs wait for review. Set `activate: true` or resume only when the user explicitly authorizes activation.
19
-
20
- Return the Job ID, status, schedules, run ID when present, and private Heny link supplied by the provider.
14
+ Return the Job ID, status, schedules, and the Agent's response. Activation and immediate runs still require the user's explicit instruction.
@@ -1,8 +1,8 @@
1
1
  ---
2
2
  name: bot-memory
3
- description: Read or manage confirmed reusable memory for a connected Bot Agent. Use when the user asks about Heny memory, Bot memory, remembered preferences, knowledge, checkpoints, or memory history.
3
+ description: Read confirmed reusable memory for a connected Bot Agent and route requested changes through the Agent's primary conversation.
4
4
  argument-hint: "[memory request]"
5
- allowed-tools: "mcp__xevy__bot_agents mcp__xevy__bot_work_start mcp__xevy__bot_memory mcp__xevy__bot_work_get mcp__xevy__bot_work_finish"
5
+ allowed-tools: "mcp__xevy__bot_agents mcp__xevy__bot_memory mcp__xevy__bot_message"
6
6
  ---
7
7
 
8
8
  # Bot memory
@@ -10,7 +10,5 @@ allowed-tools: "mcp__xevy__bot_agents mcp__xevy__bot_work_start mcp__xevy__bot_m
10
10
  Resolve the workspace and Agent. Use `bot_memory` operations:
11
11
 
12
12
  - `list`, `read`, and `history` read confirmed memory under its current private or conversation audience.
13
- - `create` runs through governed Bot work and requires scope, kind, title, and structured content.
14
- - `update` requires the memory ID plus the current `expectedVersion`; read again after a version conflict.
15
13
 
16
- Store only reusable facts the user confirms. Preserve source references and choose private scope for personal memory. Use conversation scope only when the selected conversation audience should share it. Report the saved version and audience.
14
+ Send create or update requests through `bot_message` to the Agent's primary conversation. Store only reusable facts the user confirms. Preserve source references and ask the Agent to use private scope for personal memory. Report the saved version and audience from the Agent's response.
@@ -129,20 +129,17 @@ Before saving, verify that the selected format parses, the requested content is
129
129
 
130
130
  ## 4. Save it as an Artifact
131
131
 
132
- Choose a concise kebab-case filename with the exact extension from the format table and a one-sentence description.
132
+ Choose a concise kebab-case filename with the exact extension from the format table and a one-sentence description. Save the finished Artifact to a project-local file before uploading it when filesystem access is available.
133
133
 
134
- For a project Artifact, call `create_attachment` with:
134
+ For a project Artifact, prefer file transfer so the file bytes stay outside model context and MCP JSON arguments:
135
135
 
136
- - the resolved `workspaceIdentifier` and `projectIdentifier`
137
- - exactly one content source:
138
- - `textContent` for HTML, Markdown, Mermaid, SVG, CSV, JSON, or Excalidraw
139
- - `sourceUrl` or `dataBase64` for PDF or another already-binary input
140
- - the exact `mimeType` and matching `fileName`
141
- - `description`
142
- - optional resolved `taskId`
143
- - `public: false`, unless the user explicitly requested a public/shareable Artifact
136
+ 1. When `create_attachment` exposes `filePath`, call it with the project-local path, exact `mimeType`, matching `fileName`, `description`, optional resolved `taskId`, and `public: false` unless the user explicitly requested a public/shareable Artifact. This streams files up to 100 MB.
137
+ 2. Through MCP, call `prepare_attachment_upload` with the exact filename, MIME type, byte size, description, optional task UUID, and visibility. PUT the local file bytes to the returned `uploadUrl` using the exact `uploadMethod` and `requiredHeaders`, then call `complete_attachment_upload` with the returned `attachmentId`. Completion validates the byte count, signature, and Artifact security state before the viewer URL becomes readable.
138
+ 3. Use `create_attachment` with `textContent`, `sourceUrl`, or `dataBase64` when direct local file transfer is unavailable. Keep inline content for small files; use `sourceUrl` for an existing public binary.
144
139
 
145
- For a personal-note Artifact, call `create_note_attachment` with the same content fields plus `noteId`. Note attachments are always private. Publishing a note is a separate explicit action.
140
+ Text-backed Artifacts remain limited to 10 MB after decoding for safe parsing and preview. Direct upload supports PDFs and other accepted files up to 100 MB. Never split one Artifact across calls or place a large file in base64 tool arguments.
141
+
142
+ For a personal-note Artifact, call `create_note_attachment` with the content fields plus `noteId`. Note attachments are always private. Publishing a note is a separate explicit action.
146
143
 
147
144
  ## 5. Report the result
148
145
 
@@ -1,16 +0,0 @@
1
- ---
2
- name: bot-browser
3
- description: Use the Browser assigned to a connected Bot Agent. Use whenever the user asks Heny or another Bot to browse, inspect a signed-in site, click, fill, upload, download, or perform Browser work on its Computer.
4
- argument-hint: "[browser task]"
5
- allowed-tools: "mcp__xevy__bot_agents mcp__xevy__bot_capabilities mcp__xevy__bot_work_start mcp__xevy__bot_work_heartbeat mcp__xevy__bot_tools mcp__xevy__bot_browser mcp__xevy__bot_remote mcp__xevy__bot_work_finish mcp__xevy__bot_work_cancel"
6
- ---
7
-
8
- # Bot Browser
9
-
10
- Start external work for the selected Agent, then call `bot_tools` with category `browser`. Heny exposes Browser operations as `computer.navigate`, `computer.snapshot`, `computer.read`, `computer.screenshot`, `computer.click`, `computer.fill`, `computer.select`, `computer.press`, `computer.scroll`, `computer.wait`, `computer.back`, `computer.upload`, and `computer.download`.
11
-
12
- Use only an exact returned tool and schema with `bot_browser`. Snapshot before reference-based actions and refresh after navigation or material page changes because element references are short-lived. Renew the work lease during multi-step browsing.
13
-
14
- Heny applies URL validation, Agent audience, chat-layer Decisions, and the Computer’s local policy. Inspect the action receipt before retrying an uncertain click, form submission, upload, or download. Report waiting approval with the private link.
15
-
16
- Finish the work with a concise result and evidence. Keep publication private unless explicitly requested.
@@ -1,14 +0,0 @@
1
- ---
2
- name: bot-channel
3
- description: Send through a personal Channel assigned to a connected Bot Agent. Use whenever the user asks Heny or another Bot to deliver a Telegram, email, voice, or other supported Channel message.
4
- argument-hint: "[recipient route and message]"
5
- allowed-tools: "mcp__xevy__bot_agents mcp__xevy__bot_work_start mcp__xevy__bot_tools mcp__xevy__bot_channel mcp__xevy__bot_remote mcp__xevy__bot_work_finish"
6
- ---
7
-
8
- # Bot Channels
9
-
10
- Start private external work for the Channel owner and discover category `channels`. Use an exact returned `channels.*` tool; its destination is fixed by the owner’s Heny configuration.
11
-
12
- Show the user the destination description and complete outbound content before any consequential send when Heny requests a Decision. Call `bot_channel` once with a stable key and inspect the receipt before retrying uncertain delivery.
13
-
14
- Report the provider message ID or delivery state. Channel configuration and credentials remain in Heny’s signed-in setup surface.
@@ -1,14 +0,0 @@
1
- ---
2
- name: bot-computer
3
- description: Use a Computer assigned to a connected Bot Agent for Browser, work-folder files, process execution, downloads, uploads, or locally held credentials.
4
- argument-hint: "[computer task]"
5
- allowed-tools: "mcp__xevy__bot_agents mcp__xevy__bot_capabilities mcp__xevy__bot_work_start mcp__xevy__bot_work_heartbeat mcp__xevy__bot_tools mcp__xevy__bot_computer mcp__xevy__bot_remote mcp__xevy__bot_work_finish mcp__xevy__bot_work_cancel"
6
- ---
7
-
8
- # Bot Computer
9
-
10
- Start external work and discover category `computer` with `bot_tools`. Run only exact returned `computer.*` names through `bot_computer`.
11
-
12
- Browser tools operate the Agent’s isolated persistent profile. File tools remain inside its bounded work folder. Process tools use a resolved executable and argument array with the provider timeout. Credential tools expose safe metadata and request local filling; secret values stay on the Computer.
13
-
14
- The Computer applies its own allow, deny, saved-rule, and Development-mode policy to every command. Surface local approval states accurately. Inspect receipts before retrying effects, reuse a key only for identical arguments, and finish or cancel the work when done.
@@ -1,14 +0,0 @@
1
- ---
2
- name: bot-connector
3
- description: Discover and run tools from Connectors assigned to a connected Bot Agent. Use whenever the user asks Heny or another Bot to use an assigned MCP integration or external service.
4
- argument-hint: "[connector task]"
5
- allowed-tools: "mcp__xevy__bot_agents mcp__xevy__bot_capabilities mcp__xevy__bot_work_start mcp__xevy__bot_work_heartbeat mcp__xevy__bot_tools mcp__xevy__bot_connector mcp__xevy__bot_remote mcp__xevy__bot_work_finish"
6
- ---
7
-
8
- # Bot Connector
9
-
10
- Resolve the Agent and confirm its assignments. Start external work, then call `bot_tools` with category `connectors`. Connector tools have provider-generated exact names such as `<connector>__<tool>` and may change when assignments or allow-lists change.
11
-
12
- Choose by returned description and schema. Call `bot_connector` with the exact name, arguments, and a stable key. The Connector server and Heny both enforce access and risk policy.
13
-
14
- Use `workspace.overview`, `connectors.add_mcp`, or `connectors.assign` through `bot_action` when the user asks to configure access. Let the user enter credentials through Heny’s secure setup path. Report any Decision or setup link returned by Heny.
@@ -1,17 +0,0 @@
1
- ---
2
- name: bot-skills
3
- description: List, read, create, update, enable, or run Heny product Skills through a connected Bot Agent. Use whenever the user refers to a Heny Skill, Bot Skill, reusable Bot procedure, or fixed Skill function.
4
- argument-hint: "[skill request]"
5
- allowed-tools: "mcp__xevy__bot_agents mcp__xevy__bot_work_start mcp__xevy__bot_skills mcp__xevy__bot_tools mcp__xevy__bot_skill_run mcp__xevy__bot_action mcp__xevy__bot_work_finish"
6
- ---
7
-
8
- # Heny Skills through Xevy
9
-
10
- This workflow manages Heny’s database-backed product Skills. Start external work and discover category `skills`.
11
-
12
- - Use `skills.list` to see available and active Skills.
13
- - Use `skills.use` before following or running a Skill.
14
- - Use generated `skill_<name>__run` tools for fixed functions.
15
- - Use `skills.create`, `skills.update`, and `skills.enable` through `bot_action` when requested.
16
-
17
- Follow each Skill’s returned instructions and function schema exactly. Function changes create a new version and revoke prior unattended permission. Heny applies Computer-local policy and Decisions to fixed actions. Finish with the Skill name, version, action result, and any waiting state.
@@ -1,23 +0,0 @@
1
- ---
2
- name: bot-work
3
- description: Control durable work through a connected Bot provider. Use for starting, monitoring, steering, handing off, restarting, auditing, finishing, cancelling, or attaching artifacts to Bot work.
4
- argument-hint: "[work objective or work ID]"
5
- allowed-tools: "mcp__xevy__bot_work_start mcp__xevy__bot_work_get mcp__xevy__bot_work_heartbeat mcp__xevy__bot_remote mcp__xevy__bot_tools mcp__xevy__bot_action mcp__xevy__bot_work_finish mcp__xevy__bot_work_cancel"
6
- ---
7
-
8
- # Durable Bot work
9
-
10
- Use `bot_work_start` with `controller: external` for tool-by-tool control or `bot_delegate` for native Bot execution. Every work item belongs to the connected person and preserves the Bot provider’s audience.
11
-
12
- Use `bot_remote` for the complete lifecycle:
13
-
14
- - `work_list`, `work_get`, `work_events`, and `work_receipt` inspect state and evidence.
15
- - `work_heartbeat` renews external control before the lease expires.
16
- - `work_input` supplies requested input with a stable key.
17
- - `work_handoff` transfers active external work to the native Bot Agent.
18
- - `work_restart` creates a linked attempt for terminal work.
19
- - `work_artifact` and `work_artifacts` store or list bounded durable artifacts.
20
- - `work_finish` saves the external result; publish only on explicit request.
21
- - `work_cancel` releases work that should stop.
22
-
23
- Use exact returned IDs. Preserve stable keys across retries of the same payload and create a new key when the payload changes.