agent-afk 3.89.11 → 3.90.1
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 +2 -1
- package/dist/bundled-plugins/awa-bundled/skills/shadow-verify/SKILL.md +1 -1
- package/dist/cli.mjs +332 -332
- package/dist/index.mjs +118 -118
- package/dist/telegram.mjs +137 -137
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -60,7 +60,7 @@ ANTHROPIC_API_KEY=sk-ant-...
|
|
|
60
60
|
Optional, in order of usefulness:
|
|
61
61
|
|
|
62
62
|
```bash
|
|
63
|
-
# Pick a model — opus | sonnet | haiku (Anthropic) or codex (OpenAI)
|
|
63
|
+
# Pick a model — opus | sonnet | haiku | fable (Anthropic) or codex (OpenAI)
|
|
64
64
|
AFK_MODEL=sonnet
|
|
65
65
|
|
|
66
66
|
# Enable the Telegram bot + send_telegram tool
|
|
@@ -87,6 +87,7 @@ afk chat "refactor this" --model codex
|
|
|
87
87
|
|
|
88
88
|
| Model | Best for |
|
|
89
89
|
|---|---|
|
|
90
|
+
| `fable` | Most capable — Claude Fable 5 (Mythos-class), hardest reasoning + long-horizon agentic work (1M context) |
|
|
90
91
|
| `opus` | Complex reasoning, multi-step planning, long contexts |
|
|
91
92
|
| `sonnet` | Day-to-day default — balanced speed and capability |
|
|
92
93
|
| `haiku` | Fast, cheap, one-shots |
|
|
@@ -22,7 +22,7 @@ When a sub-agent (or wave) returns investigation findings, code-review conclusio
|
|
|
22
22
|
Any time sub-agent output will drive user decisions, file edits, commits, external side-effects, or is the basis of a user-facing summary.
|
|
23
23
|
|
|
24
24
|
**Skip when:**
|
|
25
|
-
Sub-agent ran inside an orchestrator skill that already verifies (`resolve`, `diagnose`, `appmap`); sub-agent returned explicit failure; work was purely exploratory and no decision follows.
|
|
25
|
+
Sub-agent ran inside an orchestrator skill that already verifies (`resolve`, `diagnose`, `appmap`); sub-agent returned explicit failure; work was purely exploratory and no decision follows; or the session is **text-terminal** — a pure explanation, architecture walkthrough, onboarding Q&A, or capability map that names no mutated artifact (file/PR/commit/test), where there are no re-checkable state claims for adversarial verifiers to re-derive (assess coverage, coherence, and citation density instead of dispatching re-derivation sub-agents).
|
|
26
26
|
|
|
27
27
|
## Appendix: verification methods by domain (non-binding)
|
|
28
28
|
|