toga-ai 1.0.263 → 1.0.264
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/package.json +1 -1
- package/skills/kickoff/SKILL.md +50 -0
package/package.json
CHANGED
package/skills/kickoff/SKILL.md
CHANGED
|
@@ -33,6 +33,52 @@ description: Start-of-session context loader for TOGA Technology projects. Run t
|
|
|
33
33
|
> block message is never to fight it: run Steps 0–3, and the moment preflight executes you
|
|
34
34
|
> are free to read docs and work. (Emergency override only: `KICKOFF_GATE_DISABLED=1`.)
|
|
35
35
|
|
|
36
|
+
## Session permission policy — auto-accept local file I/O, ALWAYS confirm dangerous execution
|
|
37
|
+
|
|
38
|
+
> **This policy is in force for the ENTIRE session, from the moment `/kickoff` runs until it
|
|
39
|
+
> ends. It is a standing, team-wide default — apply it without being reminded.**
|
|
40
|
+
>
|
|
41
|
+
> The intent: keep the developer in flow for safe, reversible, local work, while never letting
|
|
42
|
+
> an irreversible or environment-touching action happen without an explicit human "go ahead".
|
|
43
|
+
|
|
44
|
+
**Auto-accept (proceed without a permission prompt) — safe, local, reversible:**
|
|
45
|
+
|
|
46
|
+
- **Reading files** — `Read`, `Grep`, `Glob`, and any read-only inspection of the local
|
|
47
|
+
working tree. Never pause to ask "may I read this?".
|
|
48
|
+
- **Writing/editing files** in the local working tree — `Write`, `Edit`, `NotebookEdit`, and
|
|
49
|
+
creating new local files. Proceed as long as the target is inside the developer's local
|
|
50
|
+
checkout(s). (Standard write discipline still applies — see the stub-first rule in Step 6.)
|
|
51
|
+
|
|
52
|
+
**ALWAYS ask for explicit permission first (never auto-run) — anything that executes, leaves
|
|
53
|
+
the machine, or touches a shared/remote/production environment:**
|
|
54
|
+
|
|
55
|
+
- **Shell execution** — every `Bash` / PowerShell command runner invocation. Before running
|
|
56
|
+
one, state what it does and why, and wait for the developer's "go ahead". (The kickoff gate
|
|
57
|
+
already blocks shell during priming; this extends the caution to the whole session.)
|
|
58
|
+
- **Deploys, AWS / DevOps, and infrastructure** — `/ship`, `/land-and-deploy`, `/canary`, the
|
|
59
|
+
`devops` agent's AWS CLI/console actions, Elastic Beanstalk / EC2 / RDS / S3 / IAM changes,
|
|
60
|
+
and any command that mutates infrastructure. Confirm the target environment explicitly.
|
|
61
|
+
- **Remote database access** — the `toga-db` MCP tools (`toga_query`, etc.) and any query run
|
|
62
|
+
against **production or any non-local environment**. Name the cluster/environment and get a
|
|
63
|
+
yes before executing. Treat production as require-confirmation on every call.
|
|
64
|
+
- **Network, browser, and remote git** — `WebFetch`/`WebSearch`, the `/browse` browser
|
|
65
|
+
automation, `git push`, PR creation/merge, and any call that sends data off the machine or
|
|
66
|
+
writes to a remote. Confirm before each.
|
|
67
|
+
|
|
68
|
+
**Rules of application:**
|
|
69
|
+
|
|
70
|
+
1. When in doubt about whether an action executes something or reaches outside the local
|
|
71
|
+
environment, **treat it as dangerous and ask.** The default for the unknown is *confirm*.
|
|
72
|
+
2. "Auto-accept file I/O" does **not** mean "act without judgment" — still respect
|
|
73
|
+
`/freeze`, `/guard`, and any directory scoping the developer set.
|
|
74
|
+
3. A blanket "go ahead", "just do it", or "you have full access" from the developer covers the
|
|
75
|
+
current task's expected actions, but you must **still flag** any dangerous action that falls
|
|
76
|
+
outside the scope they approved (a different environment, a production write, a deploy).
|
|
77
|
+
4. This is a **directive, not an enforcement mechanism.** For hard, mechanical enforcement the
|
|
78
|
+
team should also encode these as `permissions` allow/ask rules in the checked-in
|
|
79
|
+
`.claude/settings.json` (allow `Read`/`Edit`/`Write`; `ask` for `Bash` and the remote/DB/
|
|
80
|
+
deploy tools). Until that exists, follow this policy by discipline on every session.
|
|
81
|
+
|
|
36
82
|
## Arguments — text passed after `/kickoff` never skips any step
|
|
37
83
|
|
|
38
84
|
Trailing text (e.g. `/kickoff worker2 backend fix for Compass`, even a detailed paragraph
|
|
@@ -253,6 +299,10 @@ Tell the developer concisely:
|
|
|
253
299
|
summaries" list — the developer must see that the core is loaded and which app pulled it in.
|
|
254
300
|
- Which knowledge docs you loaded (by title). Where a repo/standard has **no knowledge yet**,
|
|
255
301
|
say so explicitly: "No knowledge captured yet for X — `capture` will build it as you work."
|
|
302
|
+
- **State the session permission policy in one line** so the developer knows the rules for
|
|
303
|
+
this session: "File reads and local edits run automatically; anything that executes, deploys,
|
|
304
|
+
hits a remote/production environment, or leaves this machine, I'll confirm with you first."
|
|
305
|
+
(See *Session permission policy* near the top of this skill.)
|
|
256
306
|
- Confirm you're primed and ready for their first task.
|
|
257
307
|
|
|
258
308
|
**Heavy-load check (from preflight `estimate`).** Preflight returns
|