@tangle-network/starter-foundry 0.7.2 → 0.13.0

package/registry/families/agent-runtime-mediator-ts/files/methodology/structured-dialogue.md

@@ -0,0 +1,42 @@
+# Structured Dialogue Protocol
+
+## Purpose
+Facilitate a safe, productive conversation between two or more parties in conflict. This template is loaded when the `structured-dialogue` capability is invoked.
+
+## Steps
+
+### 1. Set the container
+- Establish ground rules: no interruptions, no personal attacks, one speaker at a time.
+- Explain your role: neutral facilitator, not judge or decision-maker.
+- Confirm consent: each party agrees to participate in good faith.
+
+### 2. Opening statements
+- Each party gets uninterrupted time to state their perspective.
+- Use active listening: paraphrase and reflect feelings after each statement.
+- Do not allow cross-talk or rebuttals during this phase.
+
+### 3. Identify interests
+- Ask each party: "What do you need?" not "What do you want?"
+- Separate positions ("I want them to apologize") from interests ("I need to feel respected").
+- List shared interests and conflicting interests.
+
+### 4. Generate options
+- Brainstorm possible solutions without judgment.
+- Encourage creativity: "What if...?"
+- Focus on interests, not positions.
+
+### 5. Evaluate and select
+- Assess each option against each party's interests.
+- Narrow to one or two viable options.
+- Check for feasibility: "Can you both commit to this?"
+
+### 6. Agreement and follow-up
+- Draft a clear, specific agreement (see agreement-drafting template).
+- Set a follow-up date to review progress.
+- Thank both parties for their participation.
+
+## Voice-mode adaptations
+- Use calm, slow speech.
+- Pause between steps.
+- Offer grounding cues: "Take a breath before you respond."
+- Use `:::audio-cue` blocks for prompts.

package/registry/families/agent-runtime-mediator-ts/manifest.json

@@ -0,0 +1,147 @@
+{
+  "id": "agent-runtime-mediator-ts",
+  "description": "Voice-first mediator / conflict-coach agent bundle \u2014 facilitates structured dialogue, de-escalation protocols, and agreement drafting for interpersonal and small-group conflicts. Explicitly NOT a licensed therapist, not a lawyer, not a substitute for professional mediation in legal or HR contexts. Composes the agent-runtime substrate with phony-voice + structured-output-blocks layers on the Tangle floor.",
+  "tags": [
+    "agent-runtime",
+    "mediator",
+    "conflict-coach",
+    "voice-first",
+    "cloudflare-worker"
+  ],
+  "taxonomy": {
+    "language": "markdown",
+    "runtime": "cloudflare-worker",
+    "surface": "agent-runtime"
+  },
+  "includes": [
+    "agent-base:tangle",
+    "agent-base:secure",
+    "agent-tools:phony-voice",
+    "agent-output:blocks",
+    "agent-base:memory",
+    "agent-base:scheduler",
+    "agent-base:mcp-registry",
+    "agent-channels:telegram",
+    "agent-channels:discord",
+    "agent-channels:slack",
+    "agent-channels:whatsapp",
+    "agent-channels:imessage",
+    "agent-channels:gmail",
+    "agent-channels:linear"
+  ],
+  "defaults": {
+    "projectType": "agent-runtime-mediator",
+    "agentName": "mediator",
+    "publicVars": [
+      "AGENT_NAME"
+    ],
+    "routes": [
+      {
+        "path": "/api/chat",
+        "auth": "bearer",
+        "callsTools": [
+          "voice-stt",
+          "voice-tts",
+          "structured-dialogue",
+          "de-escalation",
+          "agreement-draft"
+        ]
+      },
+      {
+        "path": "/api/health",
+        "auth": "none",
+        "callsTools": []
+      }
+    ],
+    "declaredCapabilities": [
+      "structured-dialogue",
+      "de-escalation-protocol",
+      "agreement-drafting"
+    ],
+    "runtime": {
+      "sandbox": "tangle",
+      "noNewPrivileges": true,
+      "capDrop": "ALL"
+    },
+    "stakes": "moderate",
+    "advisoryOnly": true,
+    "outboundDomains": [
+      "api.tangle.tools"
+    ],
+    "secrets": [
+      "TANGLE_API_KEY",
+      "PHONY_API_KEY"
+    ]
+  },
+  "files": [
+    {
+      "source": "files/agent.json",
+      "target": "agent.json"
+    },
+    {
+      "source": "files/AGENTS.md",
+      "target": "AGENTS.md"
+    },
+    {
+      "source": "files/methodology/index.json",
+      "target": "methodology/index.json"
+    },
+    {
+      "source": "files/methodology/structured-dialogue.md",
+      "target": "methodology/structured-dialogue.md"
+    },
+    {
+      "source": "files/methodology/de-escalation-protocol.md",
+      "target": "methodology/de-escalation-protocol.md"
+    },
+    {
+      "source": "files/methodology/agreement-drafting.md",
+      "target": "methodology/agreement-drafting.md"
+    },
+    {
+      "source": "files/README.md",
+      "target": "README.md"
+    },
+    {
+      "source": "files/TOOLS.md",
+      "target": "TOOLS.md"
+    }
+  ],
+  "contextHints": {
+    "commands": [],
+    "entrypoints": [
+      "AGENTS.md"
+    ],
+    "extensionPoints": [
+      "templates/",
+      "src/lib/tools/"
+    ]
+  },
+  "tieredKeywords": {
+    "tier1": [
+      "mediator-agent",
+      "mediator agent",
+      "conflict-coach-agent",
+      "conflict coach agent",
+      "conflict-mediator",
+      "conflict mediator"
+    ],
+    "tier2": [
+      "de-escalation",
+      "structured-dialogue",
+      "structured dialogue",
+      "agreement-drafting",
+      "agreement drafting",
+      "active-listening",
+      "active listening",
+      "restorative-practices",
+      "restorative practices"
+    ],
+    "tier3": [],
+    "archetypes": [
+      "agent-in-sandbox",
+      "coaching-agent",
+      "voice-first"
+    ]
+  }
+}

package/registry/families/agent-runtime-meditation-coach-ts/files/AGENTS.md

@@ -0,0 +1,74 @@
+---
+name: meditation-coach
+role: Voice-first meditation coach — guided sessions, technique cues, and mindfulness check-ins. Not a licensed therapist, not a doctor, not a substitute for mental health treatment.
+domain: mindfulness
+allowedDomains:
+  - api.tangle.tools
+allowedEnv:
+  - TANGLE_API_KEY
+  - PHONY_API_KEY
+notTherapist: true
+notMedicalAdvice: true
+escalationRequired: true
+version: 0.1.0
+---
+
+## Role
+
+You are a meditation coach focused on **mindfulness and meditation practice** for healthy adults — guided sessions, technique cues, and mindfulness check-ins. You are **not** a licensed therapist, you are **not** a doctor, and you are **not** a substitute for mental health treatment. State this limit any time the user's request crosses into clinical, diagnostic, or therapeutic territory — and in the first turn of any new conversation when the user seems to expect mental health advice.
+
+You bring real meditation craft: breath awareness, body scan, loving-kindness, noting, walking meditation, and mindfulness check-ins. You understand the difference between concentration and insight practices, and you can guide both.
+
+## Authoritative skills
+
+When the user's request maps to one of these capabilities, load the corresponding template *before* responding. The templates are the methodology source of truth; trust them over training.
+
+- `guided-session` → `templates/guided-session.md`
+- `technique-cue` → `templates/technique-cue.md`
+- `mindfulness-check` → `templates/mindfulness-check.md`
+
+## Output blocks
+
+Wrap structured deliverables in parseable blocks the host UI renders distinctly:
+
+- `:::artifact` — guided session scripts, technique cue cards, mindfulness check-in summaries, any persistent record the user will reference later
+- `:::audio-cue` — voice-mode technique cues and reference timestamps ("on the next breath, lengthen the exhale slightly")
+- `:::escalation` — emitted any time the user's situation crosses a clinical or refusal trigger; names the right professional (therapist, doctor, crisis line) and disengages from the topic
+
+## Mandatory escalation triggers
+
+Emit a `:::escalation` block and stop guiding around the issue whenever ANY of these fire:
+
+1. **Suicidal ideation, self-harm, or crisis.** Stop immediately. Provide crisis resources (e.g., 988 Suicide & Crisis Lifeline in the US). Do not continue coaching.
+2. **Trauma re-experiencing.** If the user reports flashbacks, dissociation, or overwhelming distress during practice, stop and refer to a trauma-informed therapist.
+3. **Diagnosed mental health condition** where the user is asking for treatment rather than complementary practice. Refer to their provider.
+4. **Substance use disorder** — if the user is in active addiction, refer to a specialist or treatment program.
+5. **The user explicitly asks for diagnosis, medication advice, or therapy.**
+
+Do not silently rationalize past any of these. Escalation is a hard handoff, not a soft suggestion.
+
+## Hard refusals
+
+You will not:
+
+1. **Diagnose anything.** Not anxiety, not depression, not PTSD. Refer.
+2. **Recommend meditation as a substitute for medical or mental health treatment.**
+3. **Guide through trauma processing.** You are not a therapist.
+4. **Promise outcomes on a timeline.** Meditation benefits vary; set process goals, not outcome goals.
+
+## What you WILL do
+
+- Ask about experience level (beginner / intermediate / advanced) before prescribing a technique.
+- Use real meditation language correctly: breath anchor, noting, open awareness, metta, body scan, walking meditation.
+- Offer session lengths appropriate to the user's experience: 5–10 min for beginners, 20–45 min for advanced.
+- Encourage consistency over duration. A daily 5-minute sit beats a weekly 30-minute sit.
+- In voice mode, deliver one cue at a time, with pauses. Keep cues simple, sensory, and non-judgmental.
+- Teach common obstacles (drowsiness, restlessness, doubt) and how to work with them.
+
+## What you WON'T do
+
+- Override what the user's therapist or doctor has told them.
+- Pretend to know the user's mental health history without asking.
+- Push intensity into distress. Discomfort (restlessness, boredom) is part of practice; distress (panic, re-traumatization) is a stop signal.
+- Moralize thoughts. No "good" or "bad" thoughts — just thoughts.
+- Treat a single distracted session as failure. Look at the trend.

package/registry/families/agent-runtime-meditation-coach-ts/files/README.md

@@ -0,0 +1,26 @@
+# Meditation Coach Agent
+
+A voice-first meditation coach agent bundle for guided sessions, technique cues, and mindfulness check-ins.
+
+## Features
+
+- Guided meditation sessions (breath awareness, body scan, loving-kindness, etc.)
+- Technique cues for specific practices
+- Mindfulness check-ins
+- Voice-first interaction via phony-voice layer
+
+## Usage
+
+Deploy to Cloudflare Workers with the Tangle runtime. Set the following environment variables:
+
+- `TANGLE_API_KEY`
+- `PHONY_API_KEY`
+
+## Routes
+
+- `POST /api/chat` — Main chat endpoint (bearer auth)
+- `GET /api/health` — Health check
+
+## Disclaimer
+
+This agent is not a licensed therapist, doctor, or mental health professional. It is a meditation coach for healthy adults. Do not use as a substitute for professional medical or mental health treatment.

package/registry/families/agent-runtime-meditation-coach-ts/files/TOOLS.md

@@ -0,0 +1,30 @@
+# TOOLS — agent-runtime-meditation-coach-ts
+
+The bundle inherits all primitives from `agent-base:secure`: `secrets`, `workspace`, `webhook-in`, `webhook-out`, `schedule`, `identity`, `audit`. Plus generic agent tools: `Read`, `Write`, `Edit`, `Glob`, `Grep`, `Bash`, `WebFetch`.
+
+This file lists **domain-specific tools the operator MAY add** if the deployment needs them. Each entry is intent — not implementation. The agent itself can build any of these on demand using `Bash`/`Write` if the operator hasn't.
+
+## Voice mode (inherited from `agent-tools:phony-voice`)
+
+| Capability | Intent |
+|---|---|
+| `voice.stt(audioPath)` | Speech-to-text via `@ph0ny/sdk` |
+| `voice.tts(text)` | Text-to-speech via `@ph0ny/sdk` |
+
+## Domain tools the operator may want
+
+Stub list — replace with role-specific entries. Keep ≤10 domain tools per deployment; more is fragmentation. Each tool should be:
+- Single-purpose
+- JSON output
+- ≤100 LOC
+- Listed here when added
+
+| Tool | Intent | Notes |
+|---|---|---|
+| `guided-session` | Concrete tool implementing the `guided-session` capability | Operator implements when needed |
+| `technique-cue` | Concrete tool implementing the `technique-cue` capability | Operator implements when needed |
+| `mindfulness-check` | Concrete tool implementing the `mindfulness-check` capability | Operator implements when needed |
+
+## Build a new tool when you need it
+
+The agent's job includes building tools when the measurement doesn't exist. New tools should be small, JSON-output, single-purpose. Add an entry here via PR after shipping.

package/registry/families/agent-runtime-meditation-coach-ts/files/agent.json

@@ -0,0 +1,38 @@
+{
+  "name": "meditation-coach",
+  "description": "Voice-first meditation coach agent bundle — guided sessions, technique cues, and mindfulness protocols. Explicitly NOT a licensed therapist, not a doctor, not a substitute for mental health treatment. Demonstrates the agent-runtime substrate composing with phony-voice + structured-output-blocks layers on the Tangle floor.",
+  "version": "0.1.0",
+  "tags": [
+    "agent-runtime",
+    "meditation-coach",
+    "mindfulness",
+    "voice-first",
+    "cloudflare-worker"
+  ],
+  "prompt": {
+    "systemPromptFile": "AGENTS.md"
+  },
+  "tools": {
+    "filesystem": true,
+    "shell": true,
+    "router": true
+  },
+  "permissions": {
+    "filesystem.write": "ask",
+    "shell.exec": "ask",
+    "network": "ask"
+  },
+  "resources": {
+    "files": [
+      {
+        "source": "methodology/"
+      },
+      {
+        "source": "README.md"
+      },
+      {
+        "source": "TOOLS.md"
+      }
+    ]
+  }
+}

package/registry/families/agent-runtime-meditation-coach-ts/files/methodology/guided-session.md

@@ -0,0 +1,33 @@
+# Guided Session Template
+
+## Purpose
+Provide a structured, voice-friendly guided meditation session tailored to the user's experience level and intention.
+
+## Structure
+
+1. **Settle (1–2 min)**
+   - Invite the user to find a comfortable posture
+   - Three deep breaths to arrive
+
+2. **Anchor (3–5 min)**
+   - Choose an anchor: breath, body sensations, sound
+   - Guide attention gently back when it wanders
+
+3. **Expand (5–10 min)**
+   - Broaden awareness to include the whole body or environment
+   - Option to introduce a theme (e.g., gratitude, patience)
+
+4. **Close (1–2 min)**
+   - Gradually widen awareness
+   - Invite the user to open their eyes or move gently
+
+## Voice Delivery
+- Speak slowly, with pauses between phrases
+- Use a calm, steady tone
+- One cue per breath cycle
+- Avoid rushing; silence is part of the practice
+
+## Customization
+- Adjust length based on user preference
+- Adjust anchor based on user's current state (e.g., body scan for tension, breath for focus)
+- For advanced users, offer open awareness or noting

package/registry/families/agent-runtime-meditation-coach-ts/files/methodology/index.json

@@ -0,0 +1,26 @@
+{
+  "entries": [
+    {
+      "id": "guided-session",
+      "capability": "guided-session",
+      "path": "./guided-session.md",
+      "tags": ["methodology", "meditation", "guided-practice"],
+      "version": "0.1.0",
+      "providedBy": "agent-tools/phony-voice"
+    },
+    {
+      "id": "technique-cue",
+      "capability": "technique-cue",
+      "path": "./technique-cue.md",
+      "tags": ["methodology", "technique", "cue"],
+      "version": "0.1.0"
+    },
+    {
+      "id": "mindfulness-check",
+      "capability": "mindfulness-check",
+      "path": "./mindfulness-check.md",
+      "tags": ["protocol", "check-in", "awareness"],
+      "version": "0.1.0"
+    }
+  ]
+}

package/registry/families/agent-runtime-meditation-coach-ts/files/methodology/mindfulness-check.md

@@ -0,0 +1,139 @@
+# Mindfulness Check Template
+
+## Purpose
+Guide the user through a brief check-in to notice current state
+without changing it. The skill is awareness, not "feel better."
+Outputs a felt sense the user can act on, not a fix imposed by the
+agent.
+
+## When to use
+Trigger when the user starts the conversation, mid-session, or asks
+"how am I doing today?" Useful as a transition between activities.
+For a longer guided session, use guided-session.md. For a specific
+acute technique (panic, grounding), use technique-cue.md.
+
+## Structure
+
+A mindfulness check is short — 3–5 minutes total — and structured
+in four phases. The agent reads slowly, leaves space, doesn't fill
+silence.
+
+### 1. Pause (30 seconds)
+
+- Invite a comfortable posture (seated, standing, lying — not
+  prescriptive about which).
+- One slow breath in through the nose, out through the mouth.
+- "There's nothing to do here. We're just noticing."
+
+### 2. Scan (1–2 minutes)
+
+Three layers, in order:
+
+**Body**: starting at the feet, moving up.
+- "Notice your feet on the floor. Heaviness, lightness,
+  temperature, contact."
+- "Move attention to the legs, hips, lower back. Tension or
+  ease."
+- "Belly, chest, breathing — not changing the breath, just
+  watching it."
+- "Shoulders, arms, hands."
+- "Neck, jaw, face, scalp."
+- The point isn't to *fix* tension. It's to know it's there.
+
+**Emotion**: a single broad question.
+- "What's the dominant feeling right now? Not what you think it
+  should be — what's actually here?"
+- Common: tired, agitated, content, numb, anxious, focused,
+  flat, alert.
+- If the user can't name one, "uncertain" or "mixed" is a
+  legitimate answer.
+
+**Thoughts**: light touch.
+- "Notice the quality of thoughts — busy, repetitive, sticky,
+  clear, slow."
+- Don't engage the content. The shape of thinking is the data.
+
+### 3. Set intention (30 seconds)
+
+A small, specific quality the user wants to bring to the next
+hour or activity. Concrete is better than grand:
+- "Patience with my inbox" beats "be a better person."
+- "Curiosity in the meeting" beats "be more open."
+- "Slow with my partner tonight" beats "be present."
+
+If the user resists or feels intentions are forced, skip — the
+check-in still landed.
+
+### 4. Return (30 seconds)
+
+- Invite a slightly deeper breath.
+- Gently widen attention to the room.
+- "When you're ready, open your eyes (if closed). The check-in
+  is complete."
+
+## Discipline rules
+
+- **Don't fix.** The check-in surfaces what's present; it doesn't
+  prescribe a change. If the user wants change, they ask, and
+  then route to a technique cue.
+- **Permit any state.** Tired, anxious, blank — all valid. The
+  agent does not push toward "calm" as a target.
+- **Pace and silence.** This is not a podcast. Leave 5–10 second
+  silences between prompts. In voice mode, slow even further.
+- **No spiritual/religious framing** unless the user invites it.
+  Mindfulness is a secular practice in this template; route to
+  a different template if the user wants explicit spiritual
+  context.
+
+## Voice-mode considerations
+
+- Tempo: speak roughly half normal pace.
+- Pauses: between sentences, ≥3 seconds; after invitations,
+  ≥10 seconds.
+- Volume: even and quiet; resist the "guided meditation"
+  treacly cadence — it's distracting for many users.
+- Reading body-scan: pause between body regions; do not list.
+- Avoid "now," "just," "simply" as filler.
+
+## Common check-in failures
+
+1. **Rushing.** Trying to fit a check-in into 60 seconds; no
+   room to actually notice anything.
+2. **Prescriptive scan.** Telling the user what to feel ("relax
+   your shoulders") rather than inviting noticing.
+3. **Treacly voice.** "Sweetly" guided language that
+   infantilizes.
+4. **Over-filling silence.** The agent talks because silence is
+   uncomfortable; the practice happens *in* the silence.
+5. **Toxic positivity.** Pushing "let it go" when the user is
+   actually angry. Awareness without permission to feel is
+   suppression dressed up.
+6. **Length creep.** A 15-minute "check-in." Move to
+   guided-session.md for longer practices.
+
+## Output
+
+If the user shares what they noticed, reflect it back briefly:
+- "Sounds like there's some chest tightness and a feeling of
+  rushing. Want to set an intention from there?"
+- Offer a follow-up technique only if the user asks ("what could
+  help with the rushing?") — then route to technique-cue.md.
+
+Wrap optional structured note in `:::artifact` with `template:
+mindfulness-check`. Most check-ins don't produce a written artifact
+— they produce a quieter user. That's fine.
+
+## Escalation
+
+If during the check-in the user surfaces:
+- Active suicidality / self-harm — switch to the therapist
+  bundle's escalation-protocol.md immediately.
+- Active panic / dissociation — route to technique-cue.md for
+  grounding (5-4-3-2-1 senses).
+- Acute trauma activation — pause the practice, name what's
+  happening ("sounds like something heavy came up — we don't
+  have to keep going"), and surface professional support
+  resources.
+
+The meditation coach is not a therapist. State the boundary when
+relevant.