@quillmeetings/cli 0.1.0 → 0.1.1

package/README.md

@@ -36,7 +36,7 @@ Or run without linking:
 node bin/quill.js --help
 ```
 
-## Prerequisite: Quill MCP Bridge
+## Prerequisite: Enable Quill MCP Bridge (In Desktop App)
 
 The CLI talks to Quill through a local MCP bridge that ships with the Quill desktop app. Install Quill desktop, sign in, and enable the MCP server in Quill Settings -> MCP / Integrations.
 
@@ -204,6 +204,76 @@ Interactive browsing is disabled in agent mode. `quill browse --json` returns a
 }
 ```
 
+## LLM Agent Usage
+
+LLM agents should use `--json` for deterministic output and should avoid interactive commands.
+
+Recommended preflight:
+
+```bash
+quill doctor --json
+```
+
+Common agent-safe commands:
+
+```bash
+quill meetings list --json --limit 10
+quill meetings list --json --today
+quill search "pricing risk" --json --limit 5
+quill meetings view <meeting-id> --json
+quill notes <meeting-id> --json
+quill transcript <meeting-id> --json --full
+quill actions <meeting-id> --json --instruction "Group by owner"
+quill followup <meeting-id> --json --instruction "Tone: concise and ready to send"
+```
+
+Agent rules:
+
+- Always pass `--json` when another program will parse the result.
+- Do not call `quill browse`; it requires a real terminal.
+- If setup fails, run `quill doctor --json` and follow the first remediation.
+- Use `--fields` to reduce context for list commands.
+- Use `--full` only when the complete transcript or note body is needed.
+- Use `quill mcp tools --json` and `quill mcp schema <tool> --json` only when curated commands do not cover the task.
+
+See [SKILL.md](SKILL.md) for a compact agent instruction file that can be copied into agent systems such as local agent runners, Claude Code, or Codex.
+
+## Shell Automation
+
+Use `--json` with `jq` for scripts:
+
+```bash
+quill meetings list --json --limit 5 \
+  | jq -r '.result.result.meetings[]? | [.id, .title, .date] | @tsv'
+```
+
+Fetch notes for today's meetings:
+
+```bash
+quill meetings list --json --today \
+  | jq -r '.result.result.meetings[]?.id' \
+  | while read -r id; do
+      quill notes "$id" --json
+    done
+```
+
+Search meetings and print IDs:
+
+```bash
+quill search "customer renewal" --json --limit 10 \
+  | jq -r '.result.result.meetings[]?.id'
+```
+
+Fail fast in CI or cron:
+
+```bash
+set -euo pipefail
+quill doctor --json >/dev/null
+quill meetings list --json --limit 1 >/dev/null
+```
+
+For repeatable scripts, prefer command flags over persistent config. Precedence is: command flags > environment variables > config file > built-in defaults.
+
 ## Output
 
 Default output is human-readable. Lists render as tables in normal CLI use. Use `--format toon` for compact, YAML-like output that's cheap on LLM context, or `--json` for machine consumers.
@@ -331,7 +401,7 @@ node bin/quill.js mcp tools
 
 The test suite uses Node's built-in `node:test` and `node:assert` — no external test framework. Files live in `test/` and cover the highest-leverage code:
 
-- `test/mcp-client.test.js` — the `<ToolResponse>` XML-ish parser (`extractToolResult` / `parseQuillToolResponse`).
+- `test/mcp-client.test.js` — MCP result parsing and client error handling.
 - `test/tool-router.test.js` — fuzzy alias matching in `findTool` and key remapping in `buildArgs`.
 - `test/format.test.js` — `shapeForOutput`, `toHuman`, `toToon`, `truncateText`, etc.
 
@@ -355,13 +425,15 @@ src/cli.js            command parsing, routing, help, and MCP command wiring
 src/browser.js        Ink-powered interactive meeting browser
 src/config.js         JSON config defaults, path resolution, get/set helpers
 src/doctor.js         First-run diagnostics for Quill desktop and MCP setup
-src/mcp-client.js     Quill MCP bridge client and ToolResponse parsing
+src/mcp-client.js     Quill MCP bridge client and result parsing
 src/format.js         TOON/JSON/human output shaping, truncation, field selection
 src/tool-router.js    curated command to MCP tool mapping
 ```
 
-Implementation notes:
+## License
+
+MIT. See [LICENSE](LICENSE).
+
+## Trademarks
 
-- Quill's bridge currently returns newline-delimited JSON-RPC, not standard `Content-Length` stdio framing.
-- Quill tool results often contain XML-like `ToolResponse` text. The CLI parses the known Quill response shapes for compact display and keeps raw MCP access available for debugging.
-- MCP stdout buffers are capped by `mcp.max_buffer_bytes` to avoid unbounded memory growth on malformed or very large responses.
+Quill and Quill Meetings are trademarks of Quill Meetings. See [TRADEMARKS.md](TRADEMARKS.md).

package/SKILL.md

@@ -0,0 +1,114 @@
+# Quill CLI Agent Skill
+
+Use Quill CLI to inspect and automate Quill Meetings through the local Quill desktop MCP bridge.
+
+## Preconditions
+
+- Quill desktop must be installed and signed in.
+- The MCP server must be enabled in Quill Settings -> MCP / Integrations.
+- Run this first when setup state is unknown:
+
+```bash
+quill doctor --json
+```
+
+If `quill` is not installed globally, use:
+
+```bash
+npx @quillmeetings/cli doctor --json
+```
+
+## Agent Rules
+
+- Use `--json` for every command whose output will be parsed.
+- Do not use `quill browse`; it is interactive.
+- Prefer curated commands before raw MCP calls.
+- Use `--fields` and `--limit` to keep output small.
+- Use `--full` only when complete notes or transcripts are required.
+- If a command returns a setup or bridge error, run `quill doctor --json` and follow the first remediation.
+- Do not print or store transcript/notes output outside the user's requested destination.
+
+## Common Commands
+
+List recent meetings:
+
+```bash
+quill meetings list --json --limit 10
+```
+
+List today's meetings:
+
+```bash
+quill meetings list --json --today
+```
+
+Search meetings:
+
+```bash
+quill search "roadmap risk" --json --limit 5
+```
+
+View a meeting:
+
+```bash
+quill meetings view <meeting-id> --json
+```
+
+Read notes:
+
+```bash
+quill notes <meeting-id> --json
+```
+
+Read a full transcript:
+
+```bash
+quill transcript <meeting-id> --json --full
+```
+
+Generate action items:
+
+```bash
+quill actions <meeting-id> --json --instruction "Group by owner"
+```
+
+Generate a follow-up:
+
+```bash
+quill followup <meeting-id> --json --instruction "Tone: concise and ready to send"
+```
+
+## Raw MCP Escape Hatch
+
+Use raw MCP only when curated commands do not cover the task:
+
+```bash
+quill mcp tools --json
+quill mcp schema <tool> --json
+quill mcp call <tool> --input '{"key":"value"}' --json
+```
+
+## Shell Automation
+
+Extract meeting IDs:
+
+```bash
+quill meetings list --json --limit 10 | jq -r '.result.result.meetings[]?.id'
+```
+
+Loop over today's meetings:
+
+```bash
+quill meetings list --json --today \
+  | jq -r '.result.result.meetings[]?.id' \
+  | while read -r id; do
+      quill notes "$id" --json
+    done
+```
+
+Run a setup preflight in scripts:
+
+```bash
+set -euo pipefail
+quill doctor --json >/dev/null
+```

package/TRADEMARKS.md

@@ -0,0 +1,5 @@
+# Trademarks
+
+Quill and Quill Meetings are trademarks of Quill Meetings.
+
+This project is licensed under the MIT License, but that license does not grant permission to use Quill names, logos, or brand assets in a way that suggests endorsement, sponsorship, or affiliation beyond this CLI project.

package/package.json

@@ -1,12 +1,14 @@
 {
   "name": "@quillmeetings/cli",
-  "version": "0.1.0",
+  "version": "0.1.1",
   "description": "Human-friendly and agent-ready CLI for Quill Meetings backed by the Quill MCP server.",
   "type": "module",
   "files": [
     "bin",
     "src",
     "README.md",
+    "SKILL.md",
+    "TRADEMARKS.md",
     "LICENSE"
   ],
   "bin": {