atris 3.51.0 → 3.53.0

package/atris/skills/engines/SKILL.md

@@ -1,7 +1,7 @@
 ---
 name: engines
-description: "Dispatch work to an installed terminal agent or named Atris engine profile. Supports Atris Fast, Claude, Codex, Cursor, Fable, Composer, Haiku, Devin, Grok, and Droid. Triggers on: use codex, use cursor, use devin, use grok, use fable, use claude, use atris, engine, dispatch to, worker agent, second opinion build."
-version: 1.3.1
+description: "Dispatch work to an installed terminal agent or named Atris engine profile. Supports Atris Fast, Claude, Codex, Cursor, Fable, Composer, Haiku, Devin, Grok, and Antigravity (agy). Triggers on: use codex, use cursor, use devin, use grok, use agy, use antigravity, use fable, use claude, use atris, engine, dispatch to, worker agent, second opinion build."
+version: 1.4.0
 tags:
   - engines
   - claude
@@ -12,7 +12,8 @@ tags:
   - haiku
   - devin
   - grok
-  - droid
+  - agy
+  - antigravity
   - atris
   - orchestration
 ---
@@ -42,7 +43,7 @@ Raw spawns are not the default because they skip Atris receipts, watch, and coac
 | Haiku | `claude -p "<prompt>" --model claude-haiku-4-5` | Fast validation and bounded read-only checks. |
 | Devin | `devin -p --permission-mode dangerous -- "<prompt>"` (run from the target repo) | Default permission mode is read-only for writes — build work NEEDS `--permission-mode dangerous`, so only run it in an isolated worktree. Also `devin cloud` for sessions that outlive this machine. Supports `--model swe-1.7` |
 | Grok | `grok --always-approve -p "<prompt>"` (run from the target repo) | Headless single-turn via `-p`; default model grok-4.6. Very fast on lookups (~5-10s, reads MAP first). Great for quick second opinions; use `--best-of-n <N>` for tricky bounded builds. Uses grok.com login |
-| Droid | `droid exec "<prompt>"` | Executor profile using Droid's built-in router. Use only when the binary is ready in `atris engine --help`. |
+| Antigravity | `agy --mode accept-edits -p "<prompt>"` | `agy` executor profile. Use `--mode plan --sandbox` for read-only review and `--model <id>` to pin a model. |
 
 Headless dispatch permissions (verified 2026-08-11): `codex exec`, `grok`, and `cursor-agent` are allowlisted in `~/.claude/settings.json` so fresh and one-shot sessions can dispatch without a human approval click. A cold session that gets "requires approval" on an engine command means that allowlist regressed.
 
@@ -55,10 +56,10 @@ Headless dispatch permissions (verified 2026-08-11): `codex exec`, `grok`, and `
 - **Composer / Haiku** — fast, bounded navigation, edits, and validation where a max-tier model would be wasteful.
 - **Devin** — multi-step feature work; use `cloud` when the run should survive laptop sleep.
 - **Grok** - fastest frontier lookups and quick second opinions (grok-4.6, ~5-10s; reads MAP first); use `--best-of-n` for tricky bounded builds. Uses grok.com login.
-- **Droid** — executor work through its built-in router when installed.
+- **Antigravity / agy** — flexible executor work across Gemini, Claude, and GPT-OSS models; use `agy` as the canonical short name.
 - Parallel builds across repos: one engine job per repo, never two engines writing the same checkout.
 
-## Models worth pinning (verified live 2026-08-12)
+## Models worth pinning (verified live 2026-08-15)
 
 Each engine CLI can pin a specific model. Current best picks:
 
@@ -71,7 +72,7 @@ Each engine CLI can pin a specific model. Current best picks:
 | Haiku | `--model claude-haiku-4-5` | `haiku` for fast validation |
 | Grok | (default) | `grok-4.6` default (confirmed live 2026-08-12, ~5s lookup), `grok-4.5` still available via `-m` |
 | Codex | `-m <model>` | CLI default rides `~/.codex/config.toml`; pin with `-m` only when the task needs it |
-| Droid | (built-in router) | Let the installed Droid CLI select its model |
+| Antigravity / agy | `--model <id>` | `gemini-3.7-flash-high` for speed, `gemini-3.1-pro-high` for depth, `claude-sonnet-4-6`, `claude-opus-4-6-thinking`, or `gpt-oss-120b-medium` |
 | Atris Fast | (fixed) | api.atris.ai fast lane |
 
 Re-verify this table when a lab ships a new model: run each CLI's model-list command, smoke one lookup, and update the row. Free-tier windows (like swe-1.7 now) are the moment to fan out volume work.

package/atris/skills/youtube/SKILL.md

@@ -73,7 +73,7 @@ TOKEN=$(node -e "console.log(require('$HOME/.atris/credentials.json').token)")
 ### Process a Video
 ```bash
 atris youtube process "https://www.youtube.com/watch?v=VIDEO_ID" \
-  --query "Create a timestamped outline, claims, examples, takeaways, and action items."
+  --query "Create an outline, claims, examples, takeaways, and action items."
 ```
 
 **Parameters:**
@@ -120,13 +120,13 @@ atris youtube process "https://www.youtube.com/watch?v=..." \
 
 ### "Learn from this YouTube video"
 1. Run bootstrap
-2. Process: `atris youtube process <url> --query "Create a timestamped outline, claims, examples, takeaways, Atris implications, and next actions."`
-3. Display the timestamped analysis to the user
+2. Process: `atris youtube process <url> --query "Create an outline, claims, examples, takeaways, Atris implications, and next actions."`
+3. Display the analysis as flowing prose: ideas and who said them, never timecodes. Timestamps stay in the stored notes file for verification, not in the reply.
 
 ### "What does this video say about X?"
 1. Run bootstrap
 2. Process with focused query: `atris youtube process <url> --query "What does this say about X?"`
-3. Show the focused analysis with timestamps when available
+3. Show the focused analysis as prose; cite the speaker, not the clock
 
 ### "Process multiple videos on a topic"
 1. Run bootstrap
@@ -143,7 +143,7 @@ for url in "${VIDEOS[@]}"; do
   echo ""
 done
 ```
-3. Synthesize findings across all videos with timestamped evidence
+3. Synthesize findings across all videos; attribute ideas to speakers and videos, keep timecodes out of the reply
 
 ### "Save video insights to my agent's memory"
 1. Run bootstrap
@@ -159,7 +159,7 @@ Default output should be useful for retrieval and action:
 
 ```text
 metadata
-timestamped outline
+outline (flowing, idea-first)
 core claims with confidence
 memorable examples
 actionable takeaways
@@ -167,7 +167,7 @@ Atris/product implications
 next actions
 ```
 
-Every important insight should carry a timestamp when the transcript provides one. Treat native-video/cloud fallback output as less auditable unless it includes equivalent time anchors.
+Two layers, never mixed. The reply the person reads is flowing prose: ideas, speakers, quotes, no timecodes, nothing that reads like a stopwatch. The stored notes file keeps timestamps beside each claim so verification stays possible; that receipt layer never leaks into the reply. Treat native-video/cloud fallback output as less auditable unless the stored file includes equivalent time anchors.
 
 ## How It Works
 
@@ -205,7 +205,7 @@ npm install -g atris && atris login
 TOKEN=$(node -e "console.log(require('$HOME/.atris/credentials.json').token)")
 
 # Process a video
-atris youtube process "https://youtube.com/watch?v=..." --query "Create a timestamped outline and action brief"
+atris youtube process "https://youtube.com/watch?v=..." --query "Create a outline (flowing, idea-first) and action brief"
 
 # Process + store to agent knowledge
 atris youtube process "https://youtube.com/watch?v=..." --agent "YOUR_ID" --store